Teigland Thesis-Knowledge Networking

KNOWLEDGE NETWORKING
Structure and Performance in Networks of Practice
Robin Teigland
Structure and Performance

in Networks of Practice
Address: Holländargatan 32, P.O. Box 6501, SE-113 83 Stockholm, Sweden.
Telephone: +46 (0)8 736 9500, Fax: +46 (0)8 31 99 27, E-mail: iibve@hhs.se
• Was founded in 1975

• Is a research institute at the Stockholm School of Economics
• Conducts theoretical and empirical research with emphasis on
International Business
• Arranges weekly research seminars, yearly workshops and symposia
• Publishes its research findings and results in a series of working papers
and in published journals and books
• Is in charge of graduate courses in International Business at the
Stockholm School of Economics
• Is in charge of the International Graduate Program (IGP) at the Stockholm
School of Economics
• Is in charge of the Advanced Management Program (AMP) at the
The IIB Board of Directors
Professor Leif Lindmark, President, Stockholm School of Economics

(Chairman)
Professor Magnus Blomström, Stockholm School of Economics
Ms. Peggy Bruzelius, Chairman, Lancelot Asset Management AB
Mr. Staffan Bohman, President and CEO, Sapa AB
Mr. Jonas af Jochnick, Vice Chairman, Oriflame International S.A.
Professor Lars-Gunnar Mattsson, Stockholm School of Economics
Professor Harry Lane, Northeastern University, Boston
Mr. Anders Narvinger, CEO, The Association of Swedish Engineering
Industries
Dr. Lars Otterbeck, President and CEO, Alecta
Dr.h.c. Hans Rausing, Director (honorary member)
Mr. Tom Wachtmeister, Director
Associate Professor Peter Hagström, Director IIB
More information: www.sse.edu/iib

Further information about current research at the institute and its publications
will be mailed on request.
Structure and Performance

in Networks of Practice
Robin Teigland
Dissertation for the degree of
Doctor of Philosophy, Ph.D.

Stockholm, 2003
© IIB and the author

ISBN 91-973849-1-7
Keywords:
Knowledge
Theory of the Firm
Social Networks
Communities of Practice
Collective Action
Public Goods
Online Communities
Performance
Printed by:
Gotab, Stockholm 2003
Distributed by:
IIB, Institute of International Business
P.O. Box 6501, SE-113 83 Stockholm, Sweden.
Tel: +46 (0)8 736 9500; Fax: +46 (0)8 31 99 27; E-mail: iibve@hhs.se
To my family,
both near and far
CONTENTS
List of Figures xi
List of Tables xiii
Preface xv
Acknowledgements xvii
CHAPTER ONE
1.1 INTRODUCTION………………………………………………………….....1
1.2 DEFINING NETWORKS OF PRACTICE ............................................................. 4
1.3 OVERVIEW OF RESEARCH PURPOSES AND EMPIRICAL STUDIES .................... 6
1.3.1 Developing an Understanding of Networks of Practice ...................... 6
1.3.2 Extending Our Understanding through Empirical Studies.................. 8
1.3.3 Summary of Research Purposes......................................................... 11
1.4 DEFINITIONS AND DELIMITATIONS ............................................................. 11
1.5 PREVIEW OF THE STUDY AND MAJOR CONTRIBUTIONS .............................. 16
1.5.1 Networks of Practice .......................................................................... 16
1.5.2 Knowledge-based View of the Firm ................................................... 19
1.5.3 Practical Implications ........................................................................ 21
1.6 STRUCTURE OF THE THESIS ........................................................................ 21
CHAPTER TWO
DEVELOPING AN UNDERSTANDING OF NETWORKS OF PRACTICE
2.1 COMMUNITIES OF PRACTICE ....................................................................... 24
2.2 INTRA-ORGANIZATIONAL DISTRIBUTED NETWORKS OF PRACTICE ............. 34
2.3 INTER-ORGANIZATIONAL DISTRIBUTED NETWORKS OF PRACTICE ............. 39
2.4 ELECTRONIC NETWORKS OF PRACTICE....................................................... 48
2.5 SUMMARY .................................................................................................. 54
vii
CHAPTER THREE
REVIEW OF PREVIOUS EMPIRICAL STUDIES OF NETWORKS OF
PRACTICE
3.1 COMMUNITIES OF PRACTICE ....................................................................... 57
3.1.1 Discussion........................................................................................... 63
3.2 INTRA-ORGANIZATIONAL DISTRIBUTED NETWORKS OF PRACTICE ............. 68
3.2.1 Discussion........................................................................................... 74
3.3 INTER-ORGANIZATIONAL DISTRIBUTED NETWORKS OF PRACTICE ............. 77
3.3.1 Scientific Community Perspective...................................................... 77
3.3.2 Firm Perspective ................................................................................ 81
3.3.3 Discussion........................................................................................... 88
3.4 ELECTRONIC NETWORKS OF PRACTICE....................................................... 95
3.4.1 Discussion......................................................................................... 101
3.5 SUMMARY OF PREVIOUS EMPIRICAL STUDIES .......................................... 104
CHAPTER FOUR
DEVELOPMENT OF RESEARCH PURPOSES
4.1 RESEARCH PURPOSE 1: THE STRUCTURAL DIMENSIONS OF NETWORKS OF
PRACTICE ........................................................................................................ 113
4.1.1 RP1a: Structural Dimensions of Communities of Practice ............. 116
4.1.2 RP1b: Structural Dimensions of Electronic Networks of Practice. 117
4.2 RESEARCH PURPOSE 2: PERFORMANCE AND NETWORKS OF PRACTICE.... 120
4.3 SUMMARY OF RESEARCH PURPOSES ......................................................... 128
CHAPTER FIVE
RESEARCH METHODOLOGY
5.1 QUANTITATIVE STUDIES ........................................................................... 129
5.1.1 Internal Reliability and Validity....................................................... 130
5.2 RESEARCH METHODOLOGY FOR RESEARCH PURPOSE 1 ........................... 132
5.2.1 Research Study at Sundlink AB - Article 1....................................... 132
5.2.2 Research Study of US Professional Legal Association-Article 2..... 134
5.3 RESEARCH METHODOLOGY FOR RESEARCH PURPOSE 2 ........................... 135
5.3.1 Research Studies at Cap Gemini – Articles 4 and 5 ........................ 139
5.3.2 Research Study at Icon Medialab – Articles 3 and 6 ....................... 140
5.3.3 Research at Three High Technology Firms – Article 7 ................... 147
5.4 SUMMARY ................................................................................................ 148
5.5 GENERALIZABILITY .................................................................................. 149
viii
CHAPTER SIX
SUMMARIES OF THE EMPIRICAL STUDIES
6.0 SUMMARY OF ARTICLES ........................................................................... 151
6.1 ARTICLE 1. THEORIZING STRUCTURAL PROPERTIES OF COMMUNITIES OF
PRACTICE: A SOCIAL NETWORK APPROACH ................................................... 155
6.2 ARTICLE 2. THE PROVISION OF ONLINE PUBLIC GOODS: EXAMINING
SOCIAL STRUCTURE IN AN ELECTRONIC NETWORK OF PRACTICE ................... 163
6.3 ARTICLE 3. COMMUNITIES OF PRACTICE IN A HIGH-GROWTH INTERNET
CONSULTANCY: NETOVATION VS. ON-TIME PERFORMANCE.......................... 169
6.4 ARTICLE 4. EXTENDING RICHNESS WITH REACH: PARTICIPATION AND
KNOWLEDGE EXCHANGE IN ELECTRONIC NETWORKS OF PRACTICE ............... 175
6.5 ARTICLE 5. INTEGRATING KNOWLEDGE THROUGH INFORMATION TRADING:
EXAMINING THE IMPACT OF BOUNDARY SPANNING COMMUNICATION ON
INDIVIDUAL PERFORMANCE ............................................................................ 179
6.6 ARTICLE 6. EXPLORING THE RELATIONSHIPS BETWEEN NETWORK OF
PRACTICE PARTICIPATION, CENTRALITY, AND INDIVIDUAL PERFORMANCE IN A
MULTINATIONAL ORGANIZATION ................................................................... 183
6.7 ARTICLE 7. KNOWLEDGE DISSEMINATION IN GLOBAL R&D OPERATIONS:
AN EMPIRICAL STUDY OF MULTINATIONALS IN THE HIGH TECHNOLOGY
INDUSTRY ....................................................................................................... 189
CHAPTER SEVEN
DISCUSSION AND IMPLICATIONS
7.1 SYNTHESIS OF MAIN FINDINGS ................................................................. 193
7.1.1 Research Purpose 1: Describing Structural Dimensions ............... 193
7.1.2 Research Purpose 2: Individual Performance................................ 199
7.2 THEORETICAL IMPLICATIONS ................................................................... 208
7.2.1 Networks of Practice ......................................................................... 208
7.2.2 The Knowledge-based View of the Firm .......................................... 218
7.3 IMPLICATIONS FOR PRACTICE ................................................................... 228
7.3.1 Knowledge Management .................................................................. 228
7.3.2 Participation in Inter-organizational Networks of Practice............ 231
7.3.3 Brokers ............................................................................................. 234
7.3.4 Achieving the Balance ...................................................................... 234
7.3.5 The Future Firm ............................................................................... 235
7.4 LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH ....................... 236
7.5 CONCLUSIONS .......................................................................................... 239
REFERENCES....................................................................................241
ix
APPENDICES
APPENDIX ONE
A LOOK AT KNOWLEDGE ...............................................................265
1.1 DIMENSIONS OF KNOWLEDGE................................................................... 265
1.2 KNOWING ................................................................................................. 268
APPENDIX TWO
EMPIRICAL STUDIES .......................................................................275
ARTICLE ONE .................................................................................................... 279
ARTICLE TWO ................................................................................................... 305
ARTICLE THREE ................................................................................................ 319
ARTICLE FOUR .................................................................................................. 349
ARTICLE FIVE ................................................................................................... 363
ARTICLE SIX...................................................................................................... 393
ARTICLE SEVEN ................................................................................................ 453
x
List of Figures
Figure 1.1 Networks of Practice as a Subset of Other Networks......................... 6

Figure 1.2 Matrix of Networks of Practice .......................................................... 7
Figure 2.1 Matrix of Networks of Practice ........................................................ 24
Figure 2.2 Categories of Participation in a Community of Practice .................. 30
Figure 2.3 Distributed Network of Practice ....................................................... 35
Figure 3.1 Selected Literatures for Network of Practice Review ...................... 56
Figure 4.1 Positioning of Research Purposes in Network of Practice Matrix . 128
Figure 6.1 Positioning of Article One………………………………………..155
Figure 6.2 Technical Department..................................................................... 160
Figure 6.3 Positioning of Article Two………..………………………………163
Figure 6.4 Structure of Electronic Network of Practice................................... 167
Figure 6.5 Positioning of Article Three………………………………………169
Figure 6.6 Positioning of Article Four……………………………………….175
Figure 6.7 Positioning of Article Five………………………………………..179
Figure 6.8 Positioning of Article Six……………………...………………….183
Figure 6.9 Positioning of Article Seven…………………………...…………189
Figure 7.1 The Firm as a Social Community ................................................... 221
Figure 7.2 Absorptiveness of Ties ................................................................... 228
Figure 7.3 T-based Network of Practice Learning........................................... 231
xi
List of Tables
Table 3.1 Selected Studies of Communities of Practice .................................... 65

Table 3.2 Selected Studies of Intra-organizational Networks of Practice ......... 75
Table 3.3 Selected Studies of Inter-organizational Networks of Practice ......... 91
Table 3.4 Selected Studies of Electronic Networks of Practice....................... 102
Table 3.5 Summary of Selected Empirical Studies.......................................... 105
Table 3.6 Summary of Findings from Selected Studies on Structure.............. 107
Table 3.7 Summary of Findings from Selected Studies on Performance ........ 110
Table 4.1 Overview of Research Purposes and Corresponding Articles ......... 128
Table 5.1 Overview of Research Sites and Methods ....................................... 148
Table 6.1 Structural Properties of Communities of Practice ........................... 159
Table 6.2 Results from Stepwise Regression Analysis.................................... 171
Table 6.3 Quantitative Results from Survey .................................................... 176
Table 6.4 Results of PLS Analysis................................................................... 181
Table 7.1 Drivers of Network of Practice Centrality at Icon (Article 6) ......... 213
Table 7.2 Definitions of the Various Types of Networks of Practice.............. 215
Table 7.3 Characteristics of the Various Types of Networks of Practice ....... 216
xiii
Preface
This doctoral dissertation was written while Robin Teigland was a Ph.D.
Candidate at the Institute of International Business at the Stockholm
School of Economics. The research was in part generously funded by the
Capability Management in Network Organizations (CaMiNO)
Consortium (Askus, Ericsson, Pharmacia, SEB, Skandia, and Volvo) and
by the Xerox Corporation. This support is gratefully acknowledged.
The preparation of this thesis was made possible by the kind support and
cooperation of managers and employees of Hewlett-Packard, Cap
Gemini, and Icon Medialab, as well as of two multinational firms that
wish to remain anonymous. IIB would like to thank all the contributors
for their generosity and openness.
Additionally, IIB would like to thank McKinsey & Company for granting
Robin Teigland a leave of absence to complete her doctoral studies.
Peter Hagström
Director, Institute of International Business
xv
Acknowledgements
This thesis would not have been possible without the help and support of
many individuals, and I owe them all a huge debt of gratitude. First, each
of my committee members has made unique contributions, and I thank
them for all the time and advice they have given me throughout the
process. It has been a privilege to have had Örjan Sölvell as my thesis
chair. To Örjan I owe deep gratitude for the genuine interest he has
shown in my work and for his generous encouragement and moral
support that have not only stimulated my efforts but also helped give
direction to them. Julian Birkinshaw has been a great source of
inspiration from the very beginning. His practical suggestions and ability
to always be there whenever I needed help enabled me to get over the
many humps along the way. My many talks with Udo Zander have been
extremely insightful. His reading of the whole of my manuscript several
times has provided me with numerous and very valuable suggestions,
some of which fortunately pointed me in completely new directions. In
addition, Peter Hagström has inspired me to try to become a more
rigorous scholar – an inspiration that I will carry with me well beyond my
thesis work. Having such a diverse committee has proved very
beneficial, and I consider myself fortunate to have been able to learn from
this unique group of individuals.
In addition to my committee, I would like to especially thank two people,

Andy Schenkel and Molly Wasko. I have developed rewarding
friendships with these two individuals due to our shared hope that the
investigation of networks of practice would be valuable for organizational
researchers and practitioners. Whether over sushi in Stockholm or by
telephone from Boston, Andy has constantly questioned my thinking, yet
he has always encouraged me to stay on track while thinking about my
future. Molly and I worked together over the internet for two years
before meeting each other face-to-face, proving to me that you can trust
others and share knowledge in cyberspace without having met face-to-
xvii
face. Molly has opened my eyes to many new areas and has been a true
source of inspiration.
Additionally, so many people have been most generous with their time
through the years to help with particular problems or to read and criticize
my manuscript, either individually or in the context of group discussions.
While I cannot possibly thank everyone by name here, I do at least want
to acknowledge the following people. Over the years there has been an
outstandingly important influence of association with colleagues at IIB:
Niklas, Henrik B., Christian, Jerker, Tony, Carl, Seán, Henrik G., Gunnar,
Stefan, Ciara, Anna, Pernilla, Lin, Leonardo, Robert, Kjell, Patrick,
Jonas, Laurence, Omar, Jan-Erik, Alissa, Ivo, and Lena. Others not part
of IIB who have also provided valuable contributions, advice, and/or
support include Howard Aldrich, Hans Broström, Rob Cross, Frédéric
Delmar, Per-Olov Edlund, Caroline Haythornthwaite, Hanna Janson,
Bruce Kogut, Jan Kowalski, Håkan Ledin, Anders Lundkvist, Martha
Maznevski, Hans Pennings, Viktor Sylvan, Joachim Timlon, Cathrine
Vincenti, Gary Watson, and Barry Wellman. In particular, I am grateful
to Bill Snyder who introduced me to “communities of practice” as we sat
in a restaurant in Boston at the beginning of my doctoral studies. Bill
then opened the door to his community of practice, including Étienne
Wenger, who has continued to inspire me along the way. Additionally,
Steve Borgatti allowed me to virtually take his class on social network
analysis and has spent a great deal of time and energy helping me “get
unstuck” in the labyrinth of network analysis. A special thanks also goes
to the IIB staff, Vanja, Malin, Lotte, Anna, Kerstin, and Lena, for helping
with all those administrative issues over the years. Of note, Lotte was
particularly instrumental in the layout of this dissertation. I would also
like to thank Kerstin Ankerst and Carina Hedborg at the library for their
smiles and efficient help in gathering references throughout the years.
Several companies and individuals have played a particularly important

role in my data collection efforts. They have provided a level of support
that a doctoral student can only dream about. At Icon Medialab, I would
like to thank Usha Ananthakrishnan, Nike Barkman, Oscar Bjers, John
Ekman, Franco Fedeli, Magnus Lindahl, Jonas Petersson, Johan Staël von
Holstein, Erik Wikström, and Björn Westerberg for opening the door to
Icon and showing an interest in my research. Of special note, I am
indebted to Ulf Tingström, Marie Nyrén, and Henrik Becker for their hard
work in helping to construct and administer the social network intranet
survey at Icon. At Cap Gemini, thanks go to Carl Anlér, Patricia Evans,
xviii
Christian Forsberg, Karin Källner, and Christian Storck for providing me
with the unique opportunity to research an electronic network of practice.
Finally, I would like to thank all the individuals at Icon Medialab, Cap
Gemini, and Hewlett-Packard as well as at two anonymous multinationals
who took time out of their busy schedules to discuss with me or to
complete one of my lengthy questionnaires.
A few others deserve mentioning. While I always pondered getting a

Ph.D., it was not until I was conducting my MBA studies at Wharton that
Anjani Jain and Ananth Raman in the Decision Sciences Department
encouraged me to do so. The subsequent path to the Stockholm School of
Economics was a rather crooked one, but I would like to thank Annika
Muskantor-Bladh and Maria Öqvist, my previous colleagues at
McKinsey, for pointing me in the direction of the Institute of
International Business. I would also like to thank McKinsey for
providing me with the opportunity to take such a long leave of absence to
conduct these studies. In addition, I cannot thank enough all of my
friends in Djursholm and Stocksund who have continuously provided
moral support and have helped out with taking care of the kids (and
sometimes even the husband) as I slaved away at the computer. You
have really made Stockholm feel like home!
Finally, I would like to acknowledge my family. I am a very fortunate

person to have so many supportive individuals around me. First I would
like to thank my parents. The path to this dissertation started as a child
when I was inspired by them to explore and be curious about things. My
mother, Betsy, has always encouraged me in all my exploits, be they
rational or hot-headed, and has spent countless hours listening to me
complain, making me laugh, and offering advice. My father, Andy, who
is also an academic (organic chemist), planted the seeds for my interest in
research. When I was young, he often took me to his office where I
would play by organizing computer punch cards and building models
with molecule sets. Perhaps this is why I am so fascinated by computer
networks! My parents-in-law, Inge-Randi and Gunnar, have helped me
juggle my doctoral studies with one, two, and then three kids, along with
a hard-working husband. Additionally, I thank Bill, Bård, Elin, Kersti,
Kjell, Christine, Camilla, Lidia, Tusia, and last but definitely not least,
my brother, John, for believing in me.
On a final note, it is impossible to express what I owe those closest to me

for their dedication and untiring help, day by day. Above all, my
xix
husband, Trond, has had a special role during my time as a doctoral
student, supporting me in both good and not so good times. I thank you
Trond for your patience. And of course, I am forever grateful to my
children, Trond Gunnar, Karoline, and Aleksander, for constantly
reminding me what is important in life! “Hey kids, Mom’s book is
finally done!!!”
Stocksund, August 20, 2003
xx
CHAPTER ONE
Knowledge Networking
“..….but most importantly I have my network from the

internet. I’ve been in this for four years so really there is a
core clique of people who know each other and who trade
secrets with each other even though we have never met each
other face-to-face. We pass over nondisclosure agreements
of different companies all the time and trade company
secrets.” (Interviewee in a high technology multinational,
1998)
“People from other offices [within the company] call me up

pretty often for help because they have heard about me. But
it feels really strange to help them when you don’t know
them. It then becomes a matter of prestige - why should I
help you?” (Interviewee in a high technology multinational,
1998)
THE RECENTLY DEVELOPED knowledge-based view of the firm argues

that knowledge is the firm’s most valuable resource, yet as the above quotations
reveal, valuable firm knowledge can leak easily across the firm’s legal
boundaries while it can get stuck within them. As a result, management in
business firms is finding that knowledge cannot be “managed” using the same
tools that once were appropriate for dealing with physical goods, thus presenting
considerable challenges. A tension arises because much of the knowledge
within organizations is controlled at the level of individuals who make
discretionary choices about the sharing of knowledge. In addition to the
willingness of individuals to share knowledge, another challenge in the
management of knowledge lies in the nature of knowledge itself – that
knowledge is situated in a local practice. As a group of individuals who have a
shared practice conduct their work, boundaries are created around their practice
2 CHAPTER ONE
within which their knowledge is embedded. Within the group, tacit knowledge
is shared relatively easily between individuals often without even being made
explicit. However, sharing the group’s knowledge with others outside the group
presents difficulties even if there is a willingness to share due to the
embeddedness of the knowledge.
What is common within both these challenges to knowledge sharing is that there
is a social relationship, or lack of one, between individuals. Research has
consistently shown that social relationships are important to the ability of
individuals to gather knowledge and to perform their work (Pelz & Andrews,
1968; Mintzberg, 1973; Allen, 1977; Monge, Rothman, Eisenberg, Miller &
Kirste, 1985; Brown & Duguid, 2000; Cross, Rice, & Parker, 2001) and that the
creation of knowledge is innately a social process among individuals
(Wittgenstein, 1953; Vygotsky, 1962; Berger & Luckman, 1966). However, as
research on the work practices of individuals has revealed (cf. Wenger, 1998),
these social relationships are often not prescribed by the formal organization and
as such are “invisible”. Since individuals normally have the discretion to
interact with a range of people when they are performing their work tasks, they
form relationships based on biases and preferences for others as opposed to what
the formal organization dictates. These relationships then form the basis for
informal, naturally occurring networks that have been theoretically described as
“emergent networks” in order to distinguish them from the formally imposed or
"mandated" networks (Aldrich, 1976), which represent the legitimate authority
of an organization typically reflected by the organizational chart (Monge &
Contractor, 1997). In addition to these networks emerging within the firm,
individuals may also develop a set of emergent relationships with individuals
outside the legal boundaries of the firm, despite a lack of mandate from
management.
While organizational theorists have long discussed the importance of emergent

networks (Follett, 1924; Barnard, 1938)1, it is only recently that both scholars
and practitioners alike have showed increasing interest in them due to their
ability to serve as vital conduits of knowledge flows. If we return to the two
challenges presented in the opening paragraph of this thesis, the hope of
researchers is that an improved understanding of these emergent networks of
work-related relationships will enable firms to overcome these challenges while
facilitating their ability to create and sustain a competitive advantage. In
1
One of the first studies was by Davis (1953) who developed the “episodic communication
channels in organization” or “ecco” analysis, a technique for tracing the person-to-person
diffusion of rumors or other items of information in an organization.
KNOWLEDGE NETWORKING 3
addition, an important rationale for studying emergent networks lies in the

inconclusive findings relating formal organizational structures to organizational
behavior (Johnson, 1993; Monge & Contractor, 1997). In a review of the
empirical research on formal organizational structures, Jablin (1987) pointed out
the inconclusive nature of studies investigating an organization’s formal
structural variables, such as hierarchy, size, differentiation, and formalization.
More recently, researchers conducting a series of meta-analytic studies have
concluded that the relationships between formal structure, organizational
effectiveness (Huber, Miller, & Glick, 1990; Doty, Glick, & Huber, 1993), and
technology (Miller, Glick, Yang, & Huber, 1991) are largely an artifact of
methodological designs (Monge & Contractor, 1997). It is not surprising then
that several scholars argue that emergent structures are important to study
because they add more to our understanding of organizations than formal
organizational structures (Roethlisberger & Dickson, 1939; Roberts & O’Reilly,
1978; Bacharach & Lawler, 1980; Krackhardt & Hanson, 1993; Krikorian,
Seibold, & Goode, 1997; Monge & Contractor, 1997).
Despite the growing interest in emergent networks, we still have a very limited
understanding of them. First, while considerable research has been conducted
on emergent networks2, it is really only in recent years that researchers have
combined knowledge and learning with the study of these networks. Second,
these networks have a variety of names, such as communities of practice (Lave
& Wenger, 1991), networks of practice (Brown & Duguid, 2000), invisible
colleges (Crane, 1972), social worlds (Strauss, 1978), and scientific
communities (Polanyi, 1962b; Knorr-Cetina, 1981). However, there have been
few attempts to distinguish between the various types of networks or to review
these with a focus on structure or performance. Third, there are relatively few
empirical studies of these networks. Due to their inherent nature, these networks
are “invisible” with participants often leaving no trail of their interactions, thus
presenting a challenge to study. Not too surprisingly then, studies of these
networks tend to be of an ethnographic nature. While social network analysis
offers the possibility to study the structural dimensions of emergent networks
through making them visible, the ability to conduct such studies is, however,
often limited due to the extensive time requirements on the part of the firm as
well as the potentially “sensitive nature” of the data to be collected.
Additionally, due to their recent appearance on the scene of emergent networks,
there is extremely limited research on electronic communities, or emergent
networks in which interactions are conducted entirely online. Thus, the
overarching goal of this thesis is to improve our understanding of emergent
2
For a review, see Monge & Contractor (1997).
4 CHAPTER ONE
networks from a business firm’s perspective. Before discussing how we intend

to achieve this goal, we define the relevant concept of “network of practice” in
the next section.
1.2 Defining Networks of Practice
As seen in the previous section, there are numerous labels for networks between
individuals that emerge based upon work relationships. For the purpose of this
thesis, we will refer to the overall set of various types of emergent networks,
from communities of practice to electronic networks of practice, as networks of
practice. While this is partially in line with Brown & Duguid (2000), we also
feel that this terminology best reflects the characteristics in which we are
interested. First, the term, network, is appropriate since it implies a set of
individuals who are connected together through social relationships, whether
they are strong or weak. Terms such as community tend to denote a stronger
form of relationship, but we are interested in all networks of social relationships,
be they weak or strong. Second, we use the term practice to represent the
substrate that connects individuals in their networks (Brown & Duguid, 2001).
The principle ideas are that practice implies the actions of individuals and
groups when conducting their work, e.g., the practice of software engineers,
nurses, hotel managers, etc., and that practice involves interaction among
individuals (Lave, 1988). Thus, what distinguishes a network of practice from
other networks is that the primary reason for the emergence of relationships
within a network of practice is that individuals interact through social discourse
in order to perform their work, asking for and sharing knowledge with each
other. Thus, a network of practice can be distinguished from other networks that
emerge due to other factors, such as interests in common hobbies or discussing
sports while taking the same bus to work, etc. Finally, practice need not
necessarily be restricted to include those within one occupation or functional
discipline. Rather it may include individuals from a variety of occupations;
thus, the term, practice, is more appropriate than others such as occupation. In
summary, we define a network of practice as a set of individuals connected
together through social relationships that emerge as individuals interact on
task-related matters when conducting their work.
Before continuing to the research purposes for the empirical studies conducted
in this thesis, it is important to state our position regarding networks of practice
in relation to the formal organization. Traditionally, organizational literature has
used the distinction between formal and informal structures as a way of dividing
the interactions that occur in organizations. The formal structure has been used
to describe the organizationally specified role relationships between individuals

in formal positions (Weber, 1946; Parsons, 1951) while the informal structure
has been used to describe personal friendship relationships that often develop in
small groups (Barnard, 1938; Homans, 1950). However, several scholars
propose that the distinction between formal and informal structures is no longer
very useful (Monge & Eisenberg, 1987; Stevenson & Gilly, 1993; Monge &
Contractor, 1997) since they argue that this distinction has diminished
significantly in recent years and is expected to continue to do so (Monge &
Contractor, 1997). Reasons provided for this decline include changes to more
team-based forms of organizing, the adoption of matrix forms of organizational
structure (Burns & Wholey, 1993), shifts to network forms of organizing (Miles
& Snow, 1986, 1992, 1995; Monge, 1995), as well as the increase in lateral
communication (Galbraith, 1977) due to advances in information technologies
that enable point-to-point and broadcast communication without regard for
traditional hierarchy, distance, time, or organizational affiliation (Hinds &
Kiesler, 1995, Monge & Contractor, 1997; Faraj & Wasko, 1998).
As a result and contrary to traditional views, emergent relationships that are

ephemeral in that they are formed, maintained, broken, and reformed with
considerable ease (Palmer, Friedland, & Singh, 1986) are now argued to be the
basis for contemporary organizations (Monge & Contractor, 1997). As
Krackhardt (1994) states, "An inherent principle of the interactive form is that
networks of relations span across the entire organization, unimpeded by
preordained formal structures and fluid enough to adapt to immediate
technological demands. These relations can be multiple and complex. But one
characteristic they share is that they emerge in the organization, they are not
preplanned" (p. 218, italics in the original).
However, while these relationships do emerge, that is not to say that the formal
organization has no effect on their creation. For example, the formal
organization may bring together individuals from across the organization.
However, once the team is disbanded, individuals may continue to interact based
on their own discretion due to the building of affective bonds. While this
relationship originally is a formal one, it no longer falls under the “formal”
category. Thus, as mentioned, individuals form relationships based on biases
and preferences for others, and the creation of affective relationships may lead
them to continue to interact regardless of formally defined structures (Stevenson
& Gilly, 1993). As a result, the position on networks of practice in this thesis
falls between that of the formal organization entirely dictating interactions and
that of relationships being truly emergent since the formal structure is argued to
bias the shapes of networks of practice. Thus, in order to further clarify our
6 CHAPTER ONE
definition of networks of practice, we see networks of practice as a subset of all

potential emergent networks (e.g., friendship, common interest, etc.), and
emergent networks as a subset of all potential networks in which individuals can
participate (from formally mandated to truly emergent). We depict this
clarification in figure 1.1. Substituting now networks of practice for emergent
networks, the overarching goal of this thesis becomes to improve our
understanding of networks of practice from a business firm’s perspective. We
now turn to an overview of the means with which we intend to achieve this goal.
Figure 1.1 Networks of Practice as a Subset of Other Networks
All Networks
Emergent Networks
Networks of Practice
1.3 Overview of Research Purposes and Empirical Studies
In short, our first step is to conduct an extensive literature review to determine

our current understanding of networks of practice within and across business
firms. Based on the gaps revealed in this literature review, our next steps will be
to develop two research purposes and conduct a series of seven empirical studies
examining various networks of practice in order to fulfill these research
purposes. We discuss each of these steps in turn below.
1.3.1 Developing an Understanding of Networks of Practice

As mentioned, the first part of this thesis will focus on developing an
understanding of the various networks of practice and an overview of the
empirical studies conducted to date. In order to structure this discussion, we
develop a matrix on which we map the various types of networks of practice.
While there are many ways to define the dimensions of this matrix, we have
chosen two that we feel reflect the distinctions made in the literature to date on
networks of practice. The first dimension is the nature of the network of
practice, i.e., a set of individuals may belong to the same organization, and
within this organization they may even be co-located within the same
geographic location. The second dimension refers to the primary

communication channels used by the members of the network of practice to
interact with one another, e.g., face-to-face or totally electronically through
internet-based communication. Thus, one type of network of practice includes a
group of individuals who are co-located and who communicate primarily face-
to-face (i.e., a community of practice) while another network of practice may
include individuals from across a variety of organizations who communicate
only through the internet (i.e., an inter-organizational electronic network of
practice). Figure 1.2 provides an overview of the network of practice matrix
used to structure our discussion. Due to the fluidity of networks of practice, the
dimensions proposed here are not finite and as such there are overlaps. For
example, individuals within a community of practice may communicate both
face-to-face as well as through the company’s intranet. However, we feel that
the benefit of being able to structure our discussion based on these dimensions
outweighs any drawbacks that overlaps might cause. We would also like to note
here that this matrix is not intended to be a generic matrix that can be applied in
all situations, rather our intention is to use this matrix as a pedagogical tool with
which to structure this thesis.
Figure 1.2 Matrix of Networks of Practice
Primary Communication Channels
Face-to-face Mixed Electronic

Intra-organizational
Co-located
Nature of Network of Practice
Non-co-located
organizational
Inter-
Using this matrix, we discuss the various types of networks of practice as well
as conduct a review of the empirical studies of each type of network of practice.
This review reveals that there are two significant gaps within the areas of
8 CHAPTER ONE
structure and performance that we then use as a basis for the development of our
research purposes, as described below.
1.3.2 Extending Our Understanding through Empirical Studies

When discussing knowledge within organizations, a multitude of levels of
analysis should be taken into account and considered together (Tuomi, 1999). In
line with this and in order to increase our understanding of networks of practice
from a business firm’s perspective, the empirical studies in this thesis address
two different levels of analysis: (1) the network of practice and (2) the
individual. Each of these levels comprises a different research purpose. We
briefly present the two research purposes here while we present a more detailed
discussion of them in Chapter Four.
1.3.2.1 Research Purpose 1: Structural Dimensions of Networks of Practice

As the literature review will reveal, researchers investigating the various
networks of practice have focused primarily on understanding the cognitive
aspects relating to the interactions of individuals in networks of practice, such as
the development of a shared identity. However, studying these cognitive aspects
provides only a partial understanding of networks of practice. Within the past
few decades, researchers have been paying increasing interest to the structural
analysis of social groups. In social network theory, researchers have found that
the interactions between individuals within emergent groups create patterns of
relationships that in turn can be defined as the structure of the network (Brass,
1985; Krackhardt & Porter, 1985; Burkhardt & Brass, 1990; Krackhardt, 1991).
Some of the main principles of social network theory are that individuals are
embedded in networks of relationships that shape the patterns of behavior in
which they engage (Berkowitz, 1988), thus individuals and their actions are
viewed as interdependent rather than independent, autonomous units.
Furthermore, these network structures and the cognitive processes of individuals
in the network are argued to mutually constitute one another. However, with the
exception of a small number of studies within limited settings, researchers have
left structure by the wayside when investigating networks of practice despite
their strong parallels with the structural characteristics of embedded networks.
Thus, an application of social network measures to networks of practice should
improve our understanding of networks of practice and facilitate our ability to
further theorize and conduct empirical studies on them. For example, the
development of a set of structural properties may help to detect and analyze
communities of practice within organizations, to track their development over
time, or to measure their relationship with organizational performance. Against
the background of the above discussion, the first research purpose of this thesis
is the following:
Research Purpose 1: To describe the structural properties of networks

of practice through the application of social network analysis.
In order to address this research purpose, two studies of two polar forms of
networks of practice are conducted: an intra-organizational community of
practice and an inter-organizational electronic network of practice.
1.3.2.2 Research Purpose 2: Performance and Networks of Practice

Within recent years, researchers and practitioners alike have been increasingly
advocating networks of practice, and as a result, managers in numerous
organizations are attempting to support or even construct various forms of
networks of practice within and across their organizations (Dixon, 2000;
Wenger, McDermott, & Snyder, 2002; Ackerman, Pipek, & Wulf, 2003).
Similar to the broader set of organizational knowledge management initiatives
implemented with the purpose of enhancing performance through knowledge
sharing and creation, it seems that the hope of management is that these efforts
will positively affect individual behavior in the workplace and thus ultimately
drive increases in firm performance (Davenport & Prusak, 1998).
However, within the network of practice literature, there is a surprising scarcity

of solid academic empirical support for this generally assumed positive
relationship, with researchers paying little systematic attention to the
relationship with performance at any level. While several reasons for this lack
of rigorous investigation can be offered, this area of questioning should not go
further disregarded, as indications of a potential negative relationship between
networks of practice and performance do exist. For example, based on previous
research, it could be argued that communities of practice can evolve into core
rigidities and competency traps – inappropriate knowledge sets that preserve the
status quo and limit new insights, resulting in gaps between the knowledge of
the firm and changing market conditions (Levitt & March, 1988; Leonard-
Barton, 1992).
Additionally, as we will find in our literature review, the majority of the

research on networks of practice generally focuses on the organizational or
network level, brushing aside the fact that individuals can make choices
regarding with whom they share their knowledge and thus in which networks
they choose to participate. For example, the community of practice literature
10 CHAPTER ONE
tends to consider the already existing constellation of individuals, taking a point

of departure that individuals want to be a full member of the community of
practice and strive to be one, thus they participate in the community of practice,
freely sharing their knowledge with other members. Yet, if there is one broad
conclusion we can make from the discussion in the following chapters, it is that
individuals have a large number of networks within which they may participate
and share knowledge. Moreover, previous research investigating knowledge
sharing activity further suggests that individuals make choices regarding their
knowledge sharing, such as with whom they share their knowledge (e.g.,
Andrews & Delahaye, 2000). These choices are primarily made based upon
individual interests, and it is the variation among individuals and their interests
that results in knowledge sharing being a complicated activity often
characterized by shirking, guile, and resistance (von Krogh, 2002). For
example, theories of self-interest propose that individuals make what they
believe to be rational choices to acquire personal benefits through maximizing
(or satisficing) their gains or minimizing their losses (Monge & Contractor,
1997). Thus, individuals acting in this manner feel that there must be some
reciprocal rewards for participating in knowledge sharing, such as enjoyment,
being challenged, increased reputation, or improved performance3. In the
context of networks of practice, then we would argue that individuals make their
own decisions as to with whom they would like to interact and share knowledge
(and thus in which network of practice to participate) based on the potential
returns from these actions. While several studies have investigated individual
interests in terms of the antecedents to choices to participate within a particular
network of practice, such as increased reputation or altruism, (e.g., Lakhani &
von Hippel, 2000; Wasko & Faraj, 2000), researchers have paid scant attention
to the relationship between an individual’s participation and knowledge sharing
in a particular network and individual outcomes such as individual performance.
Tying this back to our discussion above regarding performance, we may then fill
these research gaps by bringing the level of analysis down to the individual
through investigating the relationship between an individual’s participation in
various networks of practice and individual performance. It is important to be
clear here that our research focus is not on the antecedents of these choices, such
as the how or why individuals make certain choices to participate in various
networks. Rather our interest is focused on the relationship between an
individual’s participation in various networks of practice once the choice has
3
See Monge & Contractor (1997) for a further discussion on the motives for knowledge
sharing and communication in emergent networks and Wasko (2002) for knowledge sharing
in electronic networks.
been made and individual job performance. As such, Research Purpose 2

becomes the following:
Research Purpose 2: To investigate the relationship between

individual participation in various types of networks of practice and
individual performance.
To address Research Purpose 2, we conduct a series of studies at two

multinational consulting firms that focus on individual participation in various
networks of practice and the relationship with individual performance.
1.3.3 Summary of Research Purposes

In summary, the empirical studies in this thesis have been chosen in order to fill
the considerable gaps of structure and performance in the literature on networks
of practice. Additionally, these studies serve to fill gaps in empirical research
on various types of networks of practice, such as electronic networks of practice.
As described in our discussion of knowledge in Appendix One, we argue that
knowledge is both an individual-level construct as well as a network-level
construct and so it is correctly studied at both these levels. Our approach in this
thesis then is to bring together a number of related theories that can inform our
understanding of both individual-level behaviors and network-level activities as
opposed to the organizational level. However, our intention is to then apply our
thinking and findings back to the level of the firm. Thus, by conducting a multi-
level study of various networks of practice, we may then contribute to the extant
literature on networks of practice as well as to the literature on the knowledge-
based view of the firm. While this thesis may be considered to be quite
extensive in both review and empirical studies, it does have some limitations.
These are described in the next section.
1.4 Definitions and Delimitations
The focus of this study is on networks of practice and in particular on structure

and performance in networks of practice. In this context and as defined and
discussed above, we investigate only networks of practice in this thesis and not
other networks of interaction relationships in which individuals may participate,
such as friendship networks. This is entirely in line with previous research
within the social network field that has provided evidence that friendship
networks do not impact individual outcomes such as job performance (Lazega,
2001). It is also important to note that we are only interested in networks of
practice comprised of individuals working within and across business firms. As
12 CHAPTER ONE
noted above, we define practice as an activity, an interaction among individuals

(Lave, 1988) doing their “real work”4. While practice may connote doing many
other things, such as mowing one’s own lawn or volunteering in the parent-
teacher association, we are only interested in networks of practice where the
practice revolves around the work of individuals conducted to make a living. In
connection with this point, we then limit ourselves to individuals and networks
of practice within and across business firms. In this manner, we also follow in
the footsteps of previous researchers, van Maanen & Barley (1984), who defined
an occupational community as including only those individuals who are
performing real work in order to make a living.
As for the definitions of the other main terms that we use in our thesis: structure,
performance, and knowledge, we begin with structure, the focus of our first
research purpose. We follow previous social network researchers and define
structure as the presence of regular patterns or regularities in relationships
(Wasserman & Faust, 1994) that are represented by networks comprising sets of
nodes and sets of ties depicting the interconnections between the nodes
(Wellman & Berkowitz, 1988). This definition of structure is to be
distinguished from others in the social sciences since it focuses on the “concrete
social relations among specific social actors” (ibid:5, italics in original). In
other social science approaches, while also interested in interpreting processes in
terms of patterned interrelationships among individuals rather than on the basis
of individual essences, researchers generally focus on different aspects such as
symbols, meanings, norms, values, and role expectations (ibid, Scott, 1998). In
addition, this network definition of structure is in strong contrast to the more
commonly thought of formal definition of organizational structure that refers to
the prescribed framework focusing on the differentiation of positions, the
formulation of rules and procedures, and prescriptions of authority within an
organization (Ranson, Hinings, & Greenwood, 1980)5. As mentioned above, we
are interested only in the emergent structure of networks of practice that is
determined by the patterned regularities and processes of interaction between
members as they conduct their organizational work tasks. Thus, in the context
of networks of practice, the following definition of structure will guide the
remainder of the thesis: structure is the regular patterns of relationships between
individuals (nodes) that emerge as individuals interact on task-related matters
when conducting their work (ties).
4
For a further discussion of work and practice, see Orr (1990).
5
Interest in the formal view of structure has been heavily influenced by Weber’s (1946) work
on bureaucracy. See e.g., Hall (1963), Pugh, Huckson, Hinings & Turner (1968,1969), and
Child (1972).
In our second research purpose, we move from structure to performance, and in

this thesis, we are interested only in performance in the context of a business
firm. We primarily investigate individual performance, which we refer to as the
actions or behaviors by individuals when conducting their work-related tasks
that positively influence the business firm’s goals. Thus, we follow previous
researchers who define performance not as the concrete consequences or results
of an individual’s action, e.g., monetary value of sales, but as “the action itself”
(Campbell, McCloy, Oppler, & Sager, 1992: 40). Furthermore, we view
individual performance as a multidimensional construct (Welbourne, Johnson, &
Erez, 1998). At the very least, we would expect to see a split between
“exploration” and “exploitation” (March, 1991), where exploration would be
manifested as creativity or the development of novel solutions while exploitation
would be manifested as completing one’s work-related tasks on time and on
budget. We further discuss our definition and use of performance as well as the
challenges in measuring it in Chapter Five on methodology.
Turning to knowledge, while this concept has been debated and theorized for
centuries, it continues to remain elusive. Researchers remain in disagreement as
to what knowledge is, and as a result there are numerous definitions,
overlapping terms, and perspectives regarding knowledge. (We discuss some of
these in Appendix One.) For example, while there seems to be a consensus that
a distinction needs to be made based on Polanyi’s (1962b) “knowing what” and
“knowing how”, researchers disagree as to whether the related terms, tacit and
explicit knowledge, are two distinct forms of knowledge or the ends of a
continuum (Carlile, 1997). Explicit knowledge refers to knowledge that can be
easily transmitted in formal, systematic knowledge while tacit knowledge has a
personal quality and is deeply rooted in action, commitment, and involvement in
a specific context (Nonaka, 1994). Tacit knowledge is embedded in know-how
or “the accumulated practical skill or expertise that allows one to do something
smoothly and efficiently” (von Hippel, 1988: 76), thus making it hard to
formalize and communicate (Nonaka, 1994). For our purposes here, we include
in our definition of knowledge both tacit and explicit knowledge (Nonaka,
1994). An additional characteristic of knowledge is that it enables individuals to
act in situations (von Krogh, 2002), thus keeping in line with this thesis, we
restrict our definition of knowledge to include only knowledge that enables
individuals to perform their work-related tasks.
A second major area of debate revolves around the individual/collective

dimension of knowledge. The traditional Cartesian view considers knowledge
as held by the individual since it sees the individual thinker as the primary
wielder and repository of what is known (Cook & Brown, 1999) while a more
14 CHAPTER ONE
recent sociological approach sees knowledge as socially constructed and

embedded in the social relationships between individuals (Kogut & Zander,
1992). Related to this, researchers have increasingly focused on the relationship
between the individual and organizational levels of knowledge (e.g., Hedlund &
Nonaka, 1993; Nonaka 1994; Spender, 1996); however, it is still not clear what
makes knowledge individual or organizational (Tsoukas & Vladimirou, 2001).
One view proposes that “individuals have private knowledge that can be a basis
for organizational knowledge when conveyed through speaking, gesturing,
writing, etc. Knowledge of the organization is shared knowledge among
organizational members” (von Krogh, Roos, & Slocum, 1994: 59, italics in
original). For the purposes of this thesis and compatible with this latter view,
we see individuals as having private knowledge that they share with others
through interactions revolving around work-related tasks. Additionally, it is
important to note that since individual knowledge is private, individuals are able
to make discretionary choices about the sharing of their individual knowledge,
e.g., whether to share knowledge, with whom to share the knowledge, and which
elements to disclose.
As for delimitations, despite the strengths of this thesis being a multi-level,

multi-site, multi-method study, the generalizability of the study’s discussion and
findings is an issue. First, while all the studies conducted comprise knowledge
workers, i.e., their jobs primarily consist of processing, articulating, applying
and disseminating knowledge (Wasko, 2001) as opposed to making tangible
objects with their hands, six out of the seven studies comprise individuals in
functions that are not research or science-based, e.g., researchers or scientists in
R&D labs, etc. As a result, we may not be able to generalize our findings to
networks of practice comprising research or science-based functions; however,
we may contribute to the extant literature with our studies on functions such as
lawyers, construction engineers, software programmers, behavioral scientists,
and management consultants.
Regarding the first research purpose focusing on network of practice structure,

only one community of practice and one specific type of electronic network of
practice are studied, thus limiting the generalizability of our findings to other
organizations as well as other types of electronic networks of practice. For
example, regarding electronic networks of practice, various types of interactive
technology exist such as listservs, chatrooms, and voice, and the use of these
different communication media may affect network of practice dynamics.
Regarding the second research purpose on the relationship between network of
practice participation and individual performance, we conducted our studies
addressing this purpose only in highly knowledge-intensive firms in which
individuals had access to numerous networks of practice through various

communication channels. Thus, our discussion and findings on networks of
practice within and between firms may not be generalizable to firms of a lesser
knowledge-intensive nature. However, we feel that our choice to trade breadth
for depth by focusing on deepening our understanding of networks of practice
within and across knowledge-intensive firms as opposed to a broad focus across
firms of varying knowledge intensity is more appropriate given that developed
economies have undergone a transformation from largely raw material
processing and manufacturer activities to the processing of information and
knowledge (Teece, 1998) and that knowledge work is continuing to increase as a
percentage of the total work conducted (Handy, 1991).
Finally, while our study involves individuals who perform highly knowledge-
intensive tasks and who use the new internet-based communication media to a
high degree, this study is not about the role of the internet in society, firms, or
networks of practice. We acknowledge, however, that there are two views on
the relationship between the internet and society: the technological deterministic
perspective and the social informatics perspective, and it is seemingly in order to
present these two views. Proponents of the technological deterministic
perspective tend to view the impact of the internet on society as a unilateral
process and generally fall into two camps: a utopian or a dystopian point of view
(Quan-Haase & Wellman, in press). The utopians argue that the internet will
stimulate positive change in people’s lives and work by creating new forms of
online interaction and enhancing offline relationships (cf. Sproull & Kiesler,
1991; Wellman, 2001) while the dystopians argue that the internet is fostering a
decline in social capital and an increase in interpersonal alienation (c.f. Kraut,
Patterson, Lundmark, Kiesler, Mukopadhyay, & Scherlis, 1998; Cohen, 2001).
Social informatics has arisen as a counter to the technological deterministic
perspective, with proponents arguing that the “predictions” of technological
determinists have not been fulfilled and that society is not a passive object.
Social informatics is based on “social constructivist views” of technology in
which technology emerges in dialectic with society (Quan-Haase, 2002) and has
been described as a “multidisciplinary research field that examines the design,
uses, and implications of information and communication technologies (ICTs) in
ways that account for their interactions with institutional and cultural contexts”
(Kling, 1998: 1047). In sum, in the social informatics view, neither does
technology create uses in society nor does society create uses in technology, but
they influence one another. For the purposes of clarification as to our standpoint
regarding the internet, we follow the social constructivist view due to its parallel
with similar arguments regarding knowledge and structure in the network of
practice, knowledge-based view of the firm, and social network literature.
16 CHAPTER ONE
1.5 Preview of the Study and Major Contributions
The main contribution of this thesis lies in improving our understanding of

networks of practice from a business firm’s perspective through presenting our
current understanding of networks of practice based on a literature review and
addressing two research purposes related to structure and performance through
empirical investigation. As mentioned, this empirical investigation is based on
seven separate studies. These studies, however, do complement one another,
thus this thesis provides a more comprehensive view of networks of practice and
helps advance our understanding further than if the studies had been conducted
as separate efforts by different authors. In summary, this thesis contributes to
two theoretical areas: networks of practice and the knowledge-based view of the
firm, as well as results in a number of practical implications. We discuss some
of our findings and implications below.
1.5.1 Networks of Practice

First, the extensive discussion and literature review of the various networks of
practice provide us with an understanding of the current “state of affairs” of the
network of practice field. The review of empirical studies discusses some fifty-
odd studies that we find relevant to our task at hand and summarizes these for
each type of network of practice according to three research foci: structure,
performance, and cognitive aspects. In short, this review of empirical studies
exposes that “what we think we know” is a lot more than “what we know”
regarding networks of practice. In particular, through this review we find two
significant research gaps, those of performance and structure, in addition to
finding a dearth of research relating to electronic networks of practice within
and across business firms.
Second, this thesis investigates several networks of practice in various field

settings, an important empirical contribution since networks of practice are
becoming an integral facilitator of knowledge work and new knowledge
creation. In order to address the two above research gaps of structure and
performance and as mentioned above, we develop two research purposes and
conduct a series of seven empirical studies. These studies include a variety of
research sites and data collection methods: 1) interviews and questionnaires of
members of an intra-organizational electronic network of practice in the Nordic
Operations of Cap Gemini, one of Europe’s largest IT services and management
consulting company, 2) interviews, questionnaires, and extensive sociometric
data (n=1698 in 26 offices in 16 countries spread across Europe, Asia, Australia,
and the US) in Icon Medialab, a multinational new media consulting firm, 3) a
sociometric questionnaire in Sundlink Contractors AB, a multinational

construction consortium in Scandinavia, 4) 2460 downloaded text messages and
a questionnaire in an inter-organizational electronic network of practice of
lawyers spread across the United States, and 5) case studies in the R&D
operations of three high technology multinationals, including Hewlett-Packard.
While the individuals and their networks of practice investigated are from a
variety of disciplinary and demographic backgrounds, as mentioned above, they
are all considered to be knowledge workers. To analyze the data collected, we
used a variety of methods: text analysis, multiple regression analysis, structural
equation modeling, and social network analysis, to name but a few.
As for the first research purpose, investigating structure, we synthesize social

network concepts and methods with the network of practice literature to improve
our ability to reveal these “invisible” networks through the development and
examination of structural properties of both a community of practice and an
inter-organizational electronic network of practice. Our results reveal difficulty
in applying a common set of structural properties across the board to all types of
networks of practice. Thus, we propose that the relevant structural properties
and corresponding social network measures of a particular type of network of
practice are dependent on the primary communication channels used by the
network of practice. For example, the more a network of practice depends on
electronic communication channels in which interactions are visible to all
network members (e.g., listservs, bulletin boards), the more connected members
are. Thus, social network measures such as connectedness are not as relevant in
online networks as they are in face-to-face settings. These findings regarding
structural properties then imply that the cognitive process, such as knowledge
sharing, among members vary across the different types of networks of practice.
In addition, we find support for applying theories of collective action and public
goods to electronic networks of practice and suggest that these may also be
applied to other network of practice forms to facilitate our understanding of
them.
Regarding the second research purpose, as the review of empirical studies

uncovers, we have little empirical support for the claim that there is a positive
relationship between network of practice participation and performance at any
level. In addition, most network of practice studies focus on only one network
at a time, taking the point of departure that individuals have already made their
choice to participate in the particular network. However, previous research has
indicated that individuals make choices regarding with whom they share their
knowledge. Thus, although researchers have expressed the need for a greater
understanding of participation in networks of practice, we have yet to articulate
18 CHAPTER ONE
and test a theoretical model examining the relationship between an individual’s

participation in different networks of practice and individual level outcomes.
This research is then critical for the development of a theoretical framework to
guide our understanding of the relationship between participation in different
networks of practice and individual-level outcomes. In order to address this, we
develop and test a series of hypotheses related to individual participation in
various networks in multiple research sites.
Our empirical studies reveal significant relationships between an individual’s

participation in various types of networks of practice and individual
performance. Our results suggest then that the relationship between
participation in different networks of practice and individual performance is not
only contingent upon the strength of the tie but also upon the redundancy of the
knowledge in the network. For example, we find that efficient performance has
a direct positive relationship with participation in communities of practice, yet
too much participation in communities of practice comprised of members
sharing the same functional expertise may lead to a lower degree of creative
performance. Our results also reveal that knowledge exchange and centrality are
important mediators in the relationships between participation in networks of
practice and creative performance.
Further investigation of these performance relationships reveals significant

differences between groups of individuals based on their tasks, suggesting that
the dynamics of knowledge sharing within the various networks of practice are
contingent upon the underlying practice knowledge. This finding, along with
the previous findings relating to structure and performance, supports taking a
differentiated view of networks of practice over a unitary one. Imposing one
view on networks of practice masks possible heterogeneity along two
dimensions: 1) the knowledge of the practice and 2) the form of the network of
practice; therefore, a heterogeneous view of networks of practice may be more
important in explaining outcomes than a unitary one.
Finally, we also synthesize our findings from the literature review and our
empirical studies on networks of practice and characterize the various types of
networks of practice through differences relating to their structure and
performance as well as to several other aspects such as the nature of interaction,
participation, and identity. In order to facilitate future work on networks of
practice, we also propose our own definitions of the various networks of practice
in this matrix based on our understanding of those we have developed in this
thesis.
1.5.2 Knowledge-based View of the Firm

In addition to contributing to the network of practice literature, our aim in this
thesis is to apply our thinking and empirical findings on networks of practice to
the level of the firm so that we may contribute to the extant literature on the
knowledge-based view of the firm. Turning to the knowledge-based view of the
firm literature, despite the increasing interest in this perspective, there is
interestingly limited empirical evidence to support this view of the firm.
However, we find that our research has implications for Grant’s theory of
knowledge integration, the firm as a social community (Kogut & Zander, 1992),
and the firm as a community of communities (Brown & Duguid, 1991). First,
Grant’s theory focuses primarily on the issue of coordination (structuring to
enhance the effectiveness of knowledge integration), without referring to issues
of “cooperation”. However, our empirical studies suggest that we need to
incorporate a dimension of cooperation in this view of the firm. For example, in
order for an individual to access knowledge from members in a network of
practice, our results indicate that he or she must be willing to provide knowledge
in return. Thus, norms of reciprocity and expectations of returns for knowledge
sharing appear to be key factors for participation and knowledge access in all
types of networks of practice.
Our findings are also compatible with the views of the firm as a social
community (Kogut & Zander, 1992) and a community of communities (Brown
& Duguid, 1991). At the local level, we find that individuals are members of
face-to-face communities of practice, with a high degree of participation in these
communities leading to a high degree of efficient, and in some circumstances,
creative performance. Individuals within these local communities may then also
participate to a high degree in intra-organizational distributed and electronic
networks of practice. These individuals serve as brokers, bridging local
communities of practice through exchanging, transferring, and translating
knowledge between them.
The view of the firm as a social community also argues that performance
differences among firms partly arise due to the ability of firms to transfer
knowledge within their boundaries as a result of shared identity, shared coding
schemes, shared values, and higher organizing principles (Kogut & Zander,
1992). We find suggestive evidence for this claim in our final study in which
the highest performing firm is the one that exhibits a higher degree of shared
identity across the firm as well as higher levels of knowledge sharing between
members of non-co-located intra-organizational networks of practice.
20 CHAPTER ONE
We also discuss several areas for development within the views of the firm as a
social community and community of communities. First, our studies indicate
that the issue of an individual’s membership in the firm needs to be considered.
The traditional perspective on organizational membership in the literature views
membership as a binary state. However, if the firm is to be viewed as a social
community, we should then be able to apply network of practice thinking
regarding membership to the firm as well. The argument would then be that
individuals are “members” of a firm to differing degrees. We may then
hypothesize that the degree to which an individual is a member of the firm and
in the core of the firm’s entire network of internal networks of practice will be
associated with a higher degree of individual performance. What this implicitly
argues then is that full firm members will have mastered the practice of the firm
through mutual engagement and collaboration (as predicted by theory, cf.
Wenger, 1998) and will benefit through superior performance, whereas
individuals who are less firm-like in their behavior and do not collaborate with
others in the firm will have inferior performance. If such a hypothesis is not
supported, then we have evidence that mutual engagement, collaboration, and
community membership are not valuable to performance, which would throw
into doubt the overlying argument that we see the firm as a social community.
We discuss our findings in relation to this hypothesis in the last chapter.
In relation to the above, we then discuss Kogut & Zander’s view that the firm is
a “social community of voluntaristic action” in light of the literature review and
our empirical findings. We argue again that a dimension of cooperation and
willingness to share should be incorporated in this view. Our findings suggest
that individuals weigh payoffs from participation and knowledge sharing with
other members inside the firm with payoffs from participation and knowledge
sharing with other individuals outside the firm in inter-organizational networks
of practice. Thus, in order to deepen our understanding, we propose the concept
of fuzzy individuals, borrowing from fuzzy algebra, as a means to think about
firm membership not as a binary state but as a level of degree of participation.
In addition, we discuss a new hybrid organizational form at the individual level,
in which individuals are still members of firms, but they also adopt certain
market-like behaviors as well.
Furthermore, our work on structural properties of networks of practice suggests

that we may develop a set of appropriate structural properties for viewing the
firm as a social community of which we provide some examples. These
structural properties could then facilitate further theorizing and analysis of the
knowledge-based view of the firm.
Finally, we discuss absorptive capacity (Cohen & Levinthal, 1990) and adapt
this term as an additional means to describe the ties within networks of practice
ties. In other words, we propose that while a tie between two individuals may
be weak due to infrequent communication, it may, however, be characterized by
a high level of absorptiveness. Thus, these two individuals may still be able to
share knowledge to a relatively high degree due to the ability to assimilate
knowledge through the tie. We suggest that factors affecting the absorptiveness
of a tie include the degree to which individuals have shared related previous
experience with the knowledge in question. For example, an electronic network
of practice focusing on a specific functional expertise, say C++ programming,
may be characterized by weak ties among its members, but a high degree of
absorptive ties. We would argue then that individuals participating in this
electronic network of practice, say through lurking, could still be able to
assimilate new knowledge accessed in this electronic network of practice in their
local context rather effectively due to their ability to absorb the knowledge on
account of their shared previous experience with the knowledge.
1.5.3 Practical Implications

Turning to practical contributions, given the projected growth in knowledge
management services and initiatives in the next few years, the findings of this
research have practical implications for guiding the design and maintenance of
knowledge management systems both within and between organizations. One
of the current debates in the literature is whether networks of practice can be
created and managed by a firm’s management. We do not address this debate
specifically. However, our examination of the structural dimensions of
networks of practice and the relationship between participation and individual
outcomes provides implications for how to structure social systems to support
knowledge sharing and creation within networks of practice. A set of further
practical implications is also provided.
1.6 Structure of the Thesis
This thesis is structured as follows. Chapter Two discusses the various networks
of practice using the network of practice matrix, and Chapter Three presents the
results of a review of the relevant empirical studies to date on networks of
practice. Chapter Four develops the two research purposes while Chapter Five
discusses methodological issues, providing an overview of the methods and data
collected in the seven empirical studies. Chapter Six presents a summary of the
empirical studies, and Chapter Seven concludes with a discussion of the
findings, theoretical and practical implications, limitations, and suggestions for
22 CHAPTER ONE
further research. Finally, Appendix One provides a discussion of knowledge

while Appendix Two presents the seven empirical studies in their article format.
CHAPTER TWO
Developing an Understanding of Networks

of Practice
THE PURPOSE OF this chapter is to give the reader a basic understanding of

the various networks of practice that exist both within firms as well as between
firms. In order to do so, literature from several bodies of research has been
reviewed6. We find during this review that as the literature on these networks of
practice continues to grow, so does the number of various labels for these
networks, creating a jungle of overlapping terminology. In addition, there are
no clearly agreed upon terms for the various types of networks of practice.
Thus, as mentioned in Chapter One, we will use the matrix to structure our
discussion, placing the various networks discussed in this chapter within the
matrix (figure 2.1) and labeling them in the order in which we present them.
We start our overview with a rather lengthy discussion of communities of

practice, which will then be followed by a discussion of the other networks of
practice in the matrix: intra-organizational and inter-organizational distributed
networks that are based on more traditional communication channels and then
the more recently emerging electronic networks. As can be seen, there are four
areas on the matrix that contain a “0”, indicating that we do not discuss any
networks of practice within these areas. While networks of practice within these
areas may exist, they either tend not be very common or the literature tends not
to discuss them. For example, a group of individuals co-located within one
organization generally do not communicate primarily by electronic means, thus
we have indicated a “0” in this box. Finally, while some empirical studies are
discussed in this chapter, a more thorough review of the relevant empirical
works follows in the next chapter.
6
As mentioned earlier, this thesis will only focus on networks of practice within
organizational work settings; however, research on networks of practice has been conducted
in a wide variety of settings. For example, researchers have investigated neighborhood
groups exhibiting characteristics of communities of practice that emerge around areas such as
housing, public safety, and education (e.g., Medoff & Sklar, 1994).
24 CHAPTER TWO
Figure 2.1 Matrix of Networks of Practice

Intra-organizaitonal
Co-located
1. Communities
of Practice 0/ 0/
2. Intra- 4. Intra-
Non-co-located
organizational organizational
0/ Distributed
Networks of
Electronic
Networks of
Practice Practice
3. Inter- 4. Inter-
organizational
organizational organizational
0/
Inter-
Distributed Electronic
Networks of Networks of
Practice Practice
2.1 Communities of Practice
Early research on communities of practice has its roots in situated learning

theory in ethnographic studies of work practices (Orr, 1990, 1996; Lave &
Wenger, 1991) 7. One of the most well known examples of this research is that
of Xerox service technicians in which the researcher, Julian Orr, observed that
there was a variance between the organization’s formal description of work and
the way in which the actual work was performed (Orr, 1990, 1996). When these
technicians were faced with problems for which the formal structure often did
not provide solutions, they relied on the organization’s informal systems for
help, such as story-telling, conversation, mentoring, and experiential learning.
7
There is a wider tradition in learning, education, and cognitive theory that has been
examining learning in social and situated contexts in the workplace (Abbott, 1993; Gherardi,
Nicolini, & Odella, 1998; Marsick & Watkins, 1990; Nicolini & Meznar, 1995). Studies of
managers learning in the workplace have also been conducted (Burgoyne & Hodgson, 1983;
Davies & Easterby-Smith, 1984; Fox, 1987, 1990). In addition, researchers within
psychology have begun to recognize the social and contextual dimension of learning
(Augoustinos & Walker, 1995; Farr, 1989; Gergen, 1985; Goody, 1995; Resnick, Levine, &
Teasley, 1991; Sampson, 1981).
DEVELOPING AN UNDERSTANDING 25
In so doing, these individuals formed communities of practice, a term introduced

by Lave & Wenger in 1991 and defined in the following way:
A community of practice is a set of relations among persons, activity,

and world, over time and in relation with other tangential and
overlapping communities of practice. A community of practice is an
intrinsic condition for the existence of knowledge, not least because it
provides the intrinsic support necessary for making sense of its
heritage. Thus participation in the cultural practice in which any
knowledge exists is an epistemological principle of learning. The
social structure of this practice, its power relations, and its conditions
for legitimacy define possibilities for learning (i.e., for legitimate
peripheral participation). (Lave & Wenger, 1991: 98)
What is particular to the community of practice literature is that the focus is on

learning that is situated within the work context of the relationship between
individuals and the community. Proponents of situated learning theory view
knowledge and practice as inseparable and learning as social construction.
Learning is grounded in contexts and artifacts and context is the community in
which participants learn how to conduct their work (Lave, 1988; Lave &
Wenger, 1991).
For an individual to learn and become a member of a community, individuals go

through an informal process of apprenticeship that has been labeled legitimate
peripheral participation (Lave & Wenger, 1991). In this process, individuals
learn how to function as a community member through participation in the
community, enabling them to acquire the language, values, and norms of the
community (ibid). For example, individuals learn the informal language of the
practice in social situations (Gherardi & Nicolini, 2002). This language can
only be learned through informal relationships since it is not the technical
language of the trade such as that taught in training manuals (Schenkel, 2002).
In addition, members also learn what is necessary to get the job done (Stamps,
1997) and how to handle the tasks and artifacts that they are handed (Davenport,
2002). Learning is gradually achieved as an individual moves from being a
novice, gaining access to community practices, to complete socialization and
therefore becoming an insider or full member of the community (Wenger,
1998). As individuals participate and earn their status in the community through
conducting their work, they construct and reconstruct their social identity in
relation to the community. In this manner, the legitimization process of a
community of practice differs from legitimization gained through hierarchical
26 CHAPTER TWO
status in more formal groupings such as a team. Lave & Wenger describe the
legitimization process in the following quotation:
Absorbing and being absorbed in the ‘culture of practice’ …might

include [knowing] who is involved, what they do, what everyday life is
like, how masters talk, walk, work, and generally conduct their lives,
how people who are not part of the community of practice interact
with it, what others learners are doing, and what learners need to
learn to become full practitioners. It includes an increasing
understanding of how, when, and about what old-timers collaborate,
collude, and collide, and what they enjoy, dislike, respect, and admire.
In particular it offers exemplars (which are grounds and motivation
for learning activity), including masters, finished products, and more
advanced apprentices in the process of becoming full practitioners.
(1991: 95)8.
The community of practice concept was further developed by Wenger (1998) in

which he brings together concepts of identity, meaning, practice, and
community. Following his work, communities of practice have been further
defined as emergent groups of individuals informally and contextually bound
who are applying a common competence in the pursuit of a common enterprise
(Brown & Duguid, 1991; Lave & Wenger, 1991; Snyder, 1996; Wenger, 1998).
While the term, community, is not readily definable like many widely used
terms (Etzioni, 1996), definitions generally place community in direct contrast
to society, in which self-interest, individualism, and competition reign (Tönnies,
1887; Durkheim, 1893; Weber, 1978, von Krogh 2002). One sociological
definition of community that is helpful in understanding communities of practice
is that of Bender (1982:7): “A community involves a limited number of people
in a somewhat restricted social space or network held together by shared
understandings and a sense of obligation. Relationships are close, often
intimate, and usually face-to-face. Individuals are bound together by affective
or emotional ties rather than by a perception of individual self-interest. There is
a ‘we-ness’ in a community; one is a member.” Etzioni (1996) defined
communities by two characteristics that are also prevalent in communities of
practice: 1) a web of affect-laden relationships among a group of individuals and
2) a commitment to a set of shared values, norms, and meanings, along with a
shared history and identity9.
8
Educationists are drawing heavily on Lave & Wenger’s work in trying to model the
classroom process (e.g., Fleming, 1994).
9
Other definitions of community include 1) Sarason’s (1974:1) psychological sense of
community as the sense that one is “part of a readily available, mutually supportive network
The first literature further developing communities of practice focuses on the

underlying cognitive aspects incurred in practice that serve as the source of
coherence for communities of practice. For example, Brown & Duguid (1991)
discuss three overlapping aspects of practice: collaboration, narration, and social
construction, while Wenger (1998) proposes three related characteristics of
practice: mutual engagement, joint enterprise, and shared repertoire. These
researchers see the primary requisite for the development of a community as the
mutual engagement and collaboration among individuals. As individuals engage
in the actions of their work, they collaborate with others. In the course of
everyday work, individuals may face unexpected circumstances that the
company’s formal procedures do not cover or they may be asked to conduct new
kinds of tasks, such as creating a new type of product for a customer. As
ethnographic studies have shown, individuals then engage in a fluid stream of
collaboration, helping each other to perform their tasks. For example, Wenger’s
claims processors talked and interacted as they worked while Orr’s technicians
interacted over breakfast. Often this collaboration takes the form of narration.
Through the narration of stories, individuals help each other to make sense of
circumstances that deviate from the formally described procedures. Storytelling
helps individuals to interpret events and diagnose problems through the building
of a coherent account of a random sequence of events while at the same time
developing a causal map based on their experiences. Used in this manner,
stories are more flexible than strict documentation such as that in training
manuals since they integrate contextual information, thus providing the ability to
interpret each new situation (Brown & Duguid, 1991)10. These stories then
constitute a form of collective memory for the community (Orr, 1996).
Through collaboration and narration, the members of a community of practice

negotiate meaning and joint enterprise. Joint enterprise is not a stated goal of
the community nor can it be considered to be what is defined in an individual’s
job description. Rather, it is a result of a complex process performed by a
of relationships upon which one could depend”, as well as others defining community within
a workplace setting: 2) a group of individuals who have learned how to communicate honestly
with each other, whose relationships go deeper than their masks of composure, and who have
developed a significant commitment to make others’ conditions their own (Peck, 1987), and
3) “a group of people and as a ‘way of being’ ” in which people are brought together in place
and time and barriers between people are let down (Maynard, Jr. & Mehrtens, 1993: 13). For
a discussion of the psychological sense of community in the workplace, see Klein & D’Aunno
(1986). In addition, the term community is finding its way into the management literature in
numerous ways. For example, communities for innovation was proposed by Judge, Fryxell,
& Dooley (1997) as a means to promote the innovation process in R&D operations.
10
See Schank (1995), Davenport & Prusak (1998), and Denning (2001) for a discussion of
stories as a means to capture and transfer knowledge.
28 CHAPTER TWO
community’s members in which they communally negotiate their practice and

what is necessary to perform their work while making their work habitable for
themselves (Wenger, 1998). As the community negotiates its joint enterprise,
relations of mutual accountability to the community arise as well as the
development of a common identity. In addition, the community develops a
shared repertoire of both a tacit and explicit means of communicating and
working, enabling the community to perform its practice in a satisfying manner.
The explicit includes the community’s own language and vocabulary and
artifacts such as codified procedures, documents, regulations, etc. The tacit is
the invisible side of the community, e.g., the implicit relations, cues,
unarticulated etiquette, etc. - the invisible glue that holds the community
together (Brown & Duguid, 1991; Wenger, 1998). In this manner, community
members socially construct their world based on the context in which they work,
binding themselves together by the context of the situation and creating the
social fabric of the organization in an emergent manner (Brown & Duguid,
1991).
Within a community of practice, as people interact and contribute their

knowledge to the community, trust increases among community members. As
trust increases, members become more willing to share and community
knowledge is increased. Thus, motivations to share knowledge within
communities of practice are argued to be the expectation of stronger
relationships with members as well as higher quality knowledge in future
(Davenport, 2002).
A further means of understanding communities of practice is to contrast them

with other organizational groups, such as workgroups (or teams) and social
networks. In terms of workgroups, while considerable research has been
performed on groups within the workplace, e.g., Hackman (1976), the
community of practice research differs from this research since communities of
practice are by definition emergent, self-organizing groups. This is in strong
contrast to work groups, that are generally bounded, formal organizational
entities, having been created or sanctioned by management (Brown & Duguid,
1991).
With regard to social networks, a community of practice is similar to an

informal social network of interpersonal relationships (Wellman & Berkowitz,
1988), and in fact, Wenger has noted their relationship in several places (1998:
74, 126, 287, 298). He states that communities of practice can be viewed as
nodes of “strong” ties within interpersonal networks. However, he takes pains
to point out that there is a clear distinction between the two: "A community of
practice is not defined merely by who knows whom or who talks with whom in
a network of interpersonal relations….What is of interest to me is not so much
the nature of interpersonal relationships through which information flows as the
nature of what is shared and learned and becomes a source of cohesion – that is
the structure and content of practice" (ibid: 74, 283). Thus, what distinguishes a
community of practice from other networks is that a community of practice is a
contextually based network consisting of individuals who are involved in a joint
enterprise. First, within an interpersonal network, while relationships exist
between individuals, they are not necessarily based on the pursuit of a joint
enterprise. Rather these relationships occur due to other factors, such as
common interests outside of work or discussing sports while taking the same bus
to work, etc. The ties may be based only on friendship, and the network does
not necessarily have a common goal. Within a community of practice, as
mentioned above, members share the same competence, and they use this
competence in the pursuit of a common enterprise. Thus, practice is the basis
for the community. Second, only those individuals who are seen to be
legitimate participants are included in the community, i.e., those who share the
values, language, and unwritten code of conduct. The process of becoming a
member occurs over a period of time as the individual draws upon the
community memory and collaborates with other members. Third, members
have an identity that is in relation to the community. In contrast to social
networks, individuals within communities of practice interact and collaborate in
everyday engagement and give meaning to their actions and their world through
these interactions. Membership is jointly determined and is dependent on one’s
community participation while identity involves how individuals relate to their
world and are not formed merely by being part of a social network (Wenger,
1998). Thus, every community of practice consists of a network, but not every
network forms a community of practice.
In terms of structure, Wenger discusses some structural elements of

communities of practice; however, he provides no empirical evidence of these.
Arguing that communities of practice have fluid boundaries, Wenger (1998)
proposes that there are different levels of participation within a community
(figure 2.2): (1) full participation (insider), (2) legitimate peripherality, (3)
marginality, and (4) full non-participation (outsider). Categories of participation
are not absolutes and instead are contextual and temporal, which means they are
fluid and contingent. In full participation, the person is an inclusive member of
the community. He or she has gained legitimacy through engaging with other
actors of the community in common actions and has acquired the formal and
informal ability to behave as a community member (Lave & Wenger, 1991).
The member is proficient in the tacit and explicit means of communication and
30 CHAPTER TWO
working that enable the community to perform its practice. However, legitimate
peripherality, the second category, connotes a level of only partial participation
in the community. Gaining access to the periphery is not unproblematic since
boundaries and entrance requirements may exist. For example, full participants
may develop close relationships that exclude outsiders, or a complex, detailed
understanding of the community’s practice may be required to become a full
participant. Thus, legitimate peripheral participation indicates that the
individual has gained some legitimacy among full participants. An apprentice is
one example of a legitimate peripheral participant, gaining community
knowledge and acceptance, and on his or her way to becoming a full participant.
Figure 2.2 Categories of Participation in a Community of Practice
Area of
Legitimate
Peripheral
Participation
Outsiders
Insiders
Legitimate
Peripheral
Marginal
Participants
Adapted from Tuomi (1999)
As in the case of legitimate peripherality, marginality is a mixture of

participation and non-participation. While the boundary between these two
levels is unclear, the key difference between them is the participant's trajectory
in the community. In the case of legitimate peripherality, the person is either on
an inbound trajectory to becoming a full participant or on a circular trajectory
around the periphery. However, in the case of marginality, the person's
trajectory is outbound, and he or she is thus either moving from being a full
participant to becoming an outsider or is restricted to the periphery by the
community with little hope of becoming a full participant. Marginal participants
may be best understood by looking at practices of discrimination. In such cases,
while participants wish to become community insiders, they are continuously
pushed back into identities of non-participation. Finally, the opposite of full

participation is full non-participation, or total exclusion from the community.
This form of participation may either be decided by the community or by the
non-participant since there is no desire to be part of the community.
Regarding the size of communities, discussions tend to ignore this aspect.

Wenger does argue that communities of practice are limited in size since they
are based upon feelings of mutuality and intimacy that take time to develop
through frequent interactions with others (Wenger et al., 2002), but he gives no
mention of a “ballpark” number. If we turn to research outside of the
community of practice literature, we may gain some insights. According to
anthropologists, “real” communities can rarely include more than around 150
people due to the inability of people to develop significant emotional
relationships with more than this number during the same time period. In
addition, while individuals today may know over 1000 people, they only
maintain about 20 active community ties (Kochen, 1989). This is supported by
research in non-organizational settings that suggests that North Americans
maintain an average of about 20 significant relationships at any given time
(Walker, Wasserman & Wellman, 1994). These size limits are argued to be
dictated by the “psychological preconditions for transactive memory” that have
been defined as “knowing someone well enough to know what they know, and
knowing them well enough so that you can trust them to know things in their
specialty” (Gladwell, 2000: 190). Thus, it is the recreation of the “kind of
intimacy and trust that exists in a family” (ibid).
Turning now to areas of debate within communities of practice, we find two

major areas. The first revolves around knowledge creation, innovation, and
performance, and the second discusses “management” of communities of
practice. With regard to the first debate, the first literature on communities of
practice links communities of practice positively to the creation of new
knowledge through incremental improvements in local work practices in
response to new problems (Brown & Duguid, 1991). However, recent thought
on communities of practice has also noted that while communities of practice
encourage knowledge flow and innovation within communities, they may limit
knowledge flows across communities and as such may place constraints on
innovation in the wider organization (Brown & Duguid, 2001; Swan,
Scarbrough, & Robertson, 2002). For example, some researchers argue that
innovation occurs at the interstices between established groups and work
activities since these interstices disrupt or fundamentally alter current work
practices (Blackler, 1995). Boland & Tenkasi (1995) discuss this second kind of
innovation in their work on communities of knowing, that are similar to
32 CHAPTER TWO
communities of practice yet are found in knowledge-intensive firms. Building

on work within science by Knorr-Cetina (1981) and Kuhn (1962), they argue
that it is through the dynamic interactions between communities of knowing that
new meanings and new knowledge are created in a process they describe as
perspective making and perspective taking. Perspectives are made by a
community of knowing by “refining its vocabulary, its methods, its theories and
values and its accepted logics through language and action within the
community” while perspective taking is the ability to be “able to reflect upon
and renarrativize the familiar to open up new insights and understanding”
(1995:355). Finally, in a theoretical article, Liedtka (1999) links communities of
practice with competitive advantage by proposing that organizations that are
capable of supporting the qualities of communities of practice across the
organization should be able to create and sustain a competitive advantage more
effectively than their competitors.
The second area of debate is whether communities of practice can be

constructed and managed by firm management. In response to the claims that
communities of practice are positively related to organizational performance
through innovation, efficient problem solving, best practice transfer, and the
recruitment and retention of talent (Wenger & Snyder, 2000), management
scholars have focused on ways in which organizations may exploit communities
of practice more effectively (Swan et al., 2002). Brown & Duguid (1998) argue
that organizations can play a critical role in the construction and support of
communities of practice, proposing that management can organize knowledge
across communities of practice. Recent accounts by one of the original
proponents of communities of practice, Wenger (1998, 2000a, 2000b, Wenger &
Snyder, 2000) discusses the means in which managers can construct, support,
and align communities of practice in order to exploit them for increased
innovation. However, several critics argue that this focus on control of
communities of practice for improved performance is a clear shift from
Wenger’s original interpretive approach in the seminal work with Lave (Contu
& Wilmott, 2000; Fox, 2000; Davenport, 2002). Thus, work on communities of
practice tends to fall into one of two classes: those with a more performative
perspective and those with a more interpretive or constructivist perspective
(Davenport, 2002).
One area that has not been discussed here is the “dark side” of communities of
practice. Reflecting on much of the literature to date, the above description of
communities of practice paints them primarily in a positive light. However,
Wenger has received critique for failing to discuss the wider issues of power and
conflicts that naturally occur in social relationships (Contu & Wilmott, 2000;
Fox, 2000). Research on practice prior to the community of practice literature

has revealed that resources, power, authority, legitimacy, and different
alternatives are defined and at stake within practice (Ortner, 1989; Bourdieu &
Wacquant, 1992). In addition, practice is a process of structuring (Giddens,
1979) in which individuals and groups struggle for a share of these resources,
power, authority, legitimacy, and different alternatives (Carlile, 1997). Thus, as
in any social network, individuals in central positions wield power over
resources. Newcomers to the community or organizational changes may
threaten these positions, affecting the current set of relationships. Finally, the
processes of legitimization may be no easy process. As described above,
community of practice participation is jointly determined. Thus, just because
one individual would like to become a member of a community of practice does
not necessarily mean that the individual is allowed into the community. In
addition, from the firm’s point of view, research has shown that core rigidities
and competency traps may evolve when individuals attempt to preserve the
status quo and limit new insights (Levitt & March, 1988; Leonard-Barton,
1992). Thus, communities of practice may turn into cages in which individuals
learn not to learn (Wenger, 2000).
In summary, while there are numerous versions of communities of practice

touted by both practitioners and scholars, we argue, based on our reading of the
literature, that communities of practice are generally tightly knit, emergent
groups of people who know each other relatively intimately. They primarily
work together directly in face-to-face situations since learning and knowledge
are situated within a physical setting. As such, they generally work within the
same firm, with exceptions such as consultants based at the client’s location.
Community of practice members continually and informally negotiate with,
communicate with, and coordinate with each other directly in the course of their
everyday work, and these processes are a highly implicit part of their work
practice (Brown & Duguid, 2000). Artifacts and histories are produced that aid
in the sharing of knowledge and the increase of understanding, as opposed to
achieving the performance goals such as a team has. There is a high level of
sharing of tacit and explicit knowledge among the members, and the key
characteristic of communities of practice is that the sum of a community’s
knowledge is greater than the sum of the knowledge of the individual members
(Gherardi & Nicolini, 2000). Finally, within these small groups, reciprocity is
strong, and individuals are aware of each other’s actions, resulting in a relatively
high degree of social control. As we will see, the above characteristics are in
contrast to other networks of practice that resemble more loosely coupled
systems (Weick, 1979), in which reach dominates several of these characteristics
34 CHAPTER TWO
such as reciprocity. We further discuss the differences between communities of

practice and the other networks of practice in the next section.
2.2 Intra-organizational Distributed Networks of Practice
Within an organization, communities of practice may be seen as a subset of

larger networks of practice throughout the organization. Intra-organizational
distributed networks of practice are emergent relationships of individuals who
are dispersed across the organization yet who work on similar tasks using a
similar base of knowledge and as such are to be distinguished from dispersed
teams that are formally mandated and goal-oriented11. In contrast to
communities of practice, which may comprise individuals from several different
disciplines or professions, intra-organizational distributed networks of practice
are more likely to incorporate individuals from a single discipline or profession.
The reason for this is that as individuals are separated from each other’s local
practice, the practice knowledge that they share in common declines. However,
previous research on occupational communities (e.g., van Maanen & Barley,
198412) has shown that when people work in a similar occupation, e.g., as
11
There is a growing body of research investigating dispersed teams and knowledge creation
and sharing within them. For example, the MIS literature investigates the use of information
technology enabling group processes in the context of virtual organizations, virtual
classrooms, virtual offices, virtual enterprises, and virtual teamwork. This literature focuses
primarily on group support systems (GSS), computer-mediated communication (CMC), and
electronic meeting systems (EMS) in the context of a business environment and how they
impact group communication, information sharing, and performance. See, for example,
Chidambaram & Jones (1993), Orlikowski (2002), Maznevski & Chudoba (2000), and
Sproull & Kiesler (1986, 1991). However, groups investigated in these studies often are
formally mandated by management and as such are not emergent, thus falling outside the
scope of this thesis.
12
The term occupational community has been used prior to van Maanen & Barley (1984) by
several other researchers: Gertzl (1961), Salaman (1974), Hill (1981). However, the purpose
of van Maanen & Barley’s work was to further develop the term while drawing together this
previous research as an alternative means to view behavior than through an organizational
lens. The original work on occupational communities differs from networks of practice since
the motivation for it was an understanding of organizational social control and problems of
social conflict and diversity and as such did not have knowledge and learning as the focus.
For example, within the occupational community literature, acquiring new knowledge was
important only in terms of maintaining the occupational community’s self control, and
occupational communities were not seen as a support for individuals to perform their work.
Another difference is that the literature on occupational community did not clearly specify
whether ongoing mutual engagement was a defining element of the occupational community.
Additionally, occupational communities were seen in a negative light in the original literature,
which is in direct contrast to that of the network of practice literature. However, later studies
focused on the growth of local cultures, the socialization of their members, and the
software programmers, police officers, etc., they develop similar identities,

values, and vocabularies. This shared identity and language allow people to
communicate, regardless of whether they work in the same physical location or
have a previous history of a relationship. As such, individuals from across an
organization working on a similar practice may create social networks through
which knowledge about practice can both travel rapidly and be assimilated
readily (Brown & Duguid, 2000). In this manner, intra-organizational
distributed networks of practice are similar to communities of practice in that a
shared practice is the substrate that ties members together.
Individuals who participate to a high degree in distributed networks of practice

generally serve as brokers (Wenger, 1998). These individuals act as bridges
between local communities of practice and serve to transfer and translate
knowledge between them. Due to the physically distributed nature of networks
of practice, members are generally linked together through weak ties that are not
multiplex (i.e., of only one kind, advice relationships, and are less likely to be
social). In addition, the contacts of two distributed members are less likely to
overlap the weaker the tie is between the two members (Friedkin, 1980). Figure
2.3 illustrates a distributed network of practice.
Figure 2.3 Distributed Network of Practice

Solid lines are strong ties and dotted lines are weak ties.
Distributed network of practice members: A,B,L,T
K P
O T
L
N Q
R
M S
E I
F J
C G
A B
D H
Adapted from Singer (2003)
“organization” as resulting from negotiation within communities and between external and
internal communities (Kunda, 1986; Barley, 1986; Gherardi & Nicolini, 2002).
36 CHAPTER TWO
In a distributed network of practice, coordination and communication are quite

explicit as opposed to the implicit communications of a community of practice
(Brown & Duguid, 2000). Thus, knowledge tends to be more explicit in a
distributed network of practice. Members work to a higher degree through
boundary objects or codified objects developed by communities and passed
between them, such as artifacts, texts, prescriptions, classification systems, or
indexes13. In the study of boundary objects, research has shown that the
capturing and sharing of knowledge across is no simple process (Star &
Griesemer, 1989). When knowledge is abstracted and codified, some of the tacit
knowledge is lost in the process. In addition, some aspects of locally developed
knowledge are often taken for granted and treated as common sense. As such,
individuals often have difficulty describing this knowledge or articulating its
relevance to individuals in other locations (Rennecker, 2001; Sole &
Edmondson, 2002). This inability to capture all the tacit knowledge may impact
the ability of individuals within distributed networks of practice to acquire
knowledge from distant settings and use it in their own setting. However,
despite the above, knowledge in these networks may be less redundant than that
found in member’s local communities.
The means by which intra-organizational distributed networks of practice

emerge are quite different to how a community of practice emerges since
physical closeness is not a facilitating factor. Opportunities to build practice-
based relationships across organizations may result from face-to-face meetings
such as inter-office transfers, organizational retreats or conferences, or multi-
office projects. In recent years, management in numerous companies has made
increasing attempts to facilitate the creation of relationships across their
organizations by providing the means for people working on similar tasks from
across the company to meet. The primary idea behind these actions is to
promote knowledge sharing and innovation across internal boundaries.
Examples include “technical share fairs” or “knowledge fairs” (Davenport &
Prusak, 1998) in which individuals from groups such as research, engineering,
and technical service teams are brought together to exhibit information about
their projects, expertise, and technical pursuits. In addition, management has
focused on “creating” networks of individuals from across the company,
bringing them together periodically and providing them with communication
technology, such as groupware, in order to promote interactions between the
13
Boundary objects were introduced in 1989 by Star & Griesemer, who proposed that
boundary objects provide a common ground for social actors from different social worlds to
work together. They are “plastic enough to adapt to local needs and the constraints of the
several parties employing them, yet robust enough to maintain a common identity across sites
(1989: 393).
face-to-face encounters14. These networks have numerous names such as

knowledge communities or strategic communities, and management even uses
the name community of practice for these networks, despite the inappropriate
labeling in our view. This trend has been especially prevalent in consulting
firms, such as McKinsey, which have supported the creation of networks within
areas such as transportation, banking, and finance. Traditional firms in industry
have also followed suit. For example, DaimlerChrysler has focused on bringing
together individuals who work on brakes across different product lines, e.g.,
trucks, buses, and cars. Through these different measures, individuals meet
previous acquaintances as well as make new ones, thus providing the platform
for the development of networks of practice across the organization. Thus, as
with communities of practice, the relationships that build networks of practice
can also originate due to formal measures implemented by the organization15.
Resulting from these efforts is a central area for debate as to whether a network
of practice of distributed individuals can develop into a community of practice.
Some researchers propose that communities of practice are not necessarily face-
to-face or contiguous groupings (Brown & Duguid, 1993) and as such may be
dispersed across the organization. However, we argue, based on our reading of
the community of practice literature, that communities of practice are primarily
groups of individuals involved in face-to-face interactions in co-located settings
and as such, intra-organizational distributed networks of practice cannot develop
into communities of practice. The cognitive aspects of communities of practice
described above argue for organizational locale and the resulting face-to-face
interactions in co-located sites as a significant factor in the development of
communities of practice (Sole & Edmondson, 2002). First, the ability of
dispersed individuals to develop a truly common identity in which they share the
same values and norms may be hampered by their distance (Katz & Kahn,
1966). Previous research has provided evidence that participation in different
sub-units leads to different interpretations of what is distinctive, central, and
enduring about their activities and as such leads to organizational sub-identities
(Fiol, 1991). Thus, although dispersed individuals may collaborate on activities
and negotiate meaning, their identities will more than likely differ due to local
14
In addition to these distributed networks of practice, organizations are also implementing
totally virtual networks of practice. As mentioned earlier, we will focus on networks of
practice that are based on face-to-face encounters, and we will come back to the discussion of
virtual networks of practice later in this chapter.
15
Since these communities are of a more formal nature than truly emergent, they are not the
subject of this thesis and will not be covered in more detail. Additionally, work on this type
of network tends to be more practitioner oriented, often even conducted by practitioners. For
more on these strategic communities, see for example, Storck & Hill (2000), Earl (2001),
Wenger et al. (2002), and Ackerman et al. (2003).
38 CHAPTER TWO
influences. Second, the primary processes of communities of practice involve

mutual engagement, collaboration, and narration, not merely the performance of
the same kinds of task (Brown & Duguid, 1991). These are the processes that
lead to a shared repertoire and as such, they depend on frequent interaction in
which members share experiences and recount stories often in unexpected
encounters or informal situations. These stories and experiences serve as the
community’s collective knowledge and as such guide the actions of the
community’s members, providing an understanding of the ‘way things get done
around here’. The dispersed nature of individuals within a distributed network
of practice hinders the ability of individuals to informally and frequently interact
(Kiesler & Cummings, 2002) and thus the ability to develop to the same degree
a body of communal knowledge. Third, the development of a sense of mutual
accountability to the group may also be hindered since the dispersed nature may
affect the group’s ability to develop the necessary degree of trust, commitment,
and respect (Orlikowski, 2002). Previous research has also indicated that
confidence between individuals working through electronic communications in
distributed locations has a "half-life" (de Meyer, 1991). In other words, while
confidence may be built in face-to-face relationships, once the individuals
disperse to different locations to work, the level of confidence among them
declines until they meet face-to-face again.
The final aspect regarding this debate deals with the tacit/explicit dimension of
knowledge. As described by Wenger (1998), much of the learning and
acquisition of community of practice knowledge by an individual occurs through
an implicit mode. This is in line with Reber (1993) who argues that the
acquisition of tacit knowledge takes place largely independently of conscious
attempts to learn and largely in the absence of explicit knowledge about what
was acquired. Thus, the acquisition by an individual of a community’s tacit
knowledge implies frequent interaction through word of mouth and observation,
similar to that of an apprentice and a master, and thus difficult to achieve in non-
face-to-face settings. Gherardi & Nicolini (2002) find that everyday “looking
and seeing” are two forms of action fundamental in the learning of practice.
They discuss the importance of the utterance “Look!” as important for learning
how the job is done because it signals the importance of what is happening and
the need to internalize the situation.
The importance of “looking and seeing” may differ across work practices
depending on the nature of the practice. For example, in work practices such as
construction or copy machine repair, learning and tacit knowledge are deeply
situated and distribution across physical environments hinders this process.
However, in other work practices in which the face-to-face element plays a
relatively less important role, such as software programming, learning and

knowledge sharing may occur in a more codified nature and as such may be
relatively less situated, thus facilitating the ability with which this knowledge
can be transferred through distributed networks.
Further support for communities of practice as intra-organizational face-to-face

groups in co-located settings is found if we look at the studies performed to date.
The majority of the community of practice studies are ethnographic studies of
face-to-face, intra-organizational groups in co-located settings, e.g., Suchman’s
(1983) office workers, Wenger’s (1998) claims processors, Orr’s (1990, 1996)
service technicians. Recent research has also focused on co-located settings:
Carlile’s (1997, 2002) auto supply manufacturer employees and Bechky’s
(1999, 2003) technicians and assemblers. In addition, research on whether
distributed teams can display characteristics of communities of practice also
supports our argument. For example, in a study of seven geographically
dispersed teams in a multinational company, Sole & Edmondson (2002) find
that the knowledge of the individual team members is physically situated and
thus hinders the cognitive processes identified with communities of practice
among the dispersed team members. We will return to a discussion of these and
other empirical studies in the next chapter.
In summary, intra-organizational distributed networks of practice are networks

of loosely connected individuals who tend to be members of local communities
of practice distributed across the firm. Interactions tend to be of a dyadic,
reciprocal nature, and they are conducted through both electronic and face-to-
face means. However, due to the distributed nature of the network, ties between
members are weaker than those in communities of practice and there is a lower
degree of overlap among member’s contact circles. As a result, the degree of
shared identity, language, norms, and values is also lower than in communities
of practice. While knowledge is shared through these networks, it tends to be of
a more explicit nature, often taking the form of boundary objects. Additionally,
these ties require less effort on the part of the individual to maintain and the
degree of social restraint on behavior is lower.
2.3 Inter-organizational Distributed Networks of Practice
With recent advances in information technology, the costs of informal

communication across a firm’s legal boundaries by individuals throughout
hierarchical levels and functional competence groups in the organization have
been greatly reduced (DeSanctis & Fulk, 1999; Kettinger & Grover, 1997).
40 CHAPTER TWO
Thus, individuals may easily access individuals across the firm’s legal
boundaries (Cronin & Rosenbaum, 1994) and discuss their work tasks with
professional contacts, friends, family, ex-colleagues, or other acquaintances who
are working on similar tasks. We refer to these networks of individuals working
on similar tasks yet dispersed across organizations as inter-organizational
networks of practice.
For individuals not working within a highly professional capacity or at higher

levels of the hierarchy, previous efforts to interact with others outside their
organizations were often fruitless as they could be time-consuming or against
company policies unless specified in the job description. Indeed, individuals
may not even have known whom to contact or how to find a relevant person.
Furthermore, if management did not provide the resources to attend external
conferences or other events, finding other like-minded individuals with whom to
discuss work-related problems often proved difficult. Since these
communication developments are recent, we know little about inter-
organizational networks of practice other than those comprising individuals who
are primarily engaged in work with a highly professional content, such as
researchers within science-based industries. We now turn to this research that
has been conducted within this capacity.
Inter-organizational networks of practice are by no means a new phenomenon.

They have existed for hundreds of years and have played an important role in
the diffusion of knowledge through society. For example, in the history of
science, “gentlemen scientists” belonging to scientific communities sent letters
back and forth to enable individuals to keep up-to-date with developments as
well as to establish some form of control over the development of the
community’s knowledge (Pinch, 1990). In addition, they fall under a variety of
names and are characterized by varying degrees of connectedness in terms of
relationships: scientific communities (Polanyi, 1962b; Knorr-Cetina, 1981), co-
citation networks (Usdiken & Pasadeos, 1995), invisible colleges (Crane, 1972),
epistemic communities (Holzner & Marx, 1979; Adler & Haas, 1992; Haas,
1992), thought-collectives (Fleck, 1935), paradigms (Kuhn, 1962), and
occupational communities (van Maanen & Barley, 1984)16. The term, invisible
16
Another well-known inter-organizational grouping is social worlds (Strauss 1978). Social
worlds are one of the broadest groupings of individuals, including groups ranging from
scientists to stamp collectors to baseball fans. Social worlds are fluid groups of individuals
whose glue is based upon a central activity (and potentially related activities). However,
since the activities of a social world in their general sense could be hobbies such as stamp
collecting, baseball, or politics, they are not necessarily related to an individual’s practice or
work. While individuals within social worlds are described as developing their identity in
colleges, dates back to the 1640s when a group of ten men, who actually were
not professional scientists but were well-educated within one field, would meet
informally in the taverns of London. These meetings later developed in 1660
into the Royal Society, the oldest scientific society in Great Britain (Price, 1963;
Tuire & Erno, 2001). Since then, invisible colleges have been used to describe
groups of researchers within the same branch of science who have personal
relationships with one another (Crane, 1972). The other above terms, however,
tend to have a broader definition of membership. For example, van Maanen &
Barley suggest that “professions are viewed as occupational communities”
(1984: 287), i.e., boundaries are determined by the profession and not by
personal relationships. Despite their broader definitions, we include these in our
discussion since research on these sheds some light on our investigation of inter-
organizational networks of practice. Finally, in order to avoid repetition, we
will use some of these terms interchangeably.
Research on occupational communities has provided evidence that individuals

see their work not only as a means of making a living, but as a “central life
interest” (Dubin, 1956), deriving meaning and value from their work while
constructing their social identity in relation to the occupational community (van
Maanen & Barley, 1984). Members share values, norms, and perspectives, and
since they are often widely dispersed they maintain their relationships through
conferences, conventions, newsletters, publications, and other means. In
addition, norms and behaviors are enforced through the use of professional
associations to accredit educational programs in their profession, to prevent
individuals who are not sanctioned from participating, and to discipline
individuals for misconduct or malpractice (Pickering & King, 1995). The
majority of research on these inter-organizational networks has been conducted
on networks of individuals who are involved in the physical or social sciences
working as researchers, thus the following discussion reflects this bias17.
relation to the social world, such as in a network of practice, there are some primary
differences between social worlds and networks of practice. First, research on social worlds
has tended not to focus on knowledge and learning. Second, membership in social worlds is
generally determined merely by affiliation and is not necessarily determined by the greater
group through social interaction. Third, the world is held together through communication
and not social interaction since communication may be one way. For example, an individual
may affiliate and identify himself with the social world of baseball because he likes to watch
baseball on television. However, he does not necessarily have to interact with anyone about
baseball, nor may he be interested in learning more about the sport or improving it. As a
result, this literature provides little understanding of networks of practice that are non-science
based.
17
While much of the research on science-based networks has been conducted within
academic settings, we have chosen to include this body of literature since these networks and
42 CHAPTER TWO
Regarding knowledge sharing and creation within these inter-organizational

networks, individuals engage in debate and discussion of each other’s ideas and
results as well as collaborate on joint research projects (Crane, 1972; Ziman,
1978)18. In many cases, these informal relationships are more valuable than
publications for sharing knowledge since the results of failed experiments are
rarely published and learning about these can prevent their duplication. Through
these relationships, knowledge and innovations spread across national and
cultural boundaries. Due to the universal nature of the knowledge within the
community as well as a shared language and values, individuals can
communicate relatively easily with one another across these boundaries
(Tushman & Katz, 1980; van Maanen & Barley, 1984). In addition to
presentations and authorship of publications (Liebeskind, Oliver, Zucker, &
Brewer, 1996), personal communication is an effective means of establishing
and protecting individuals’ intellectual property rights (Price, 1963).
Knowledge is often considered to be more of a public good than a private good
within the community since the publishing of results makes knowledge available
to all within the community.
In addition, the central goals and values of a community are developed and
spread throughout the networks (Hagstrom, 1965). Strong norms that are well
defined and socially enforced exist within scientific communities, such as
reciprocity in knowledge sharing, respect for individuals’ intellectual property
rights, and honesty in research (Crane, 1972; Blau, 1973; Liesbeskind et al.,
1996; Debackere & Rappa, 1994; Bouty, 2000). Trustworthy behavior and
norms are enforced since the level of participation in the community is jointly
determined by the community’s members. Individuals who fail to follow the
norms and implicit code of conduct can be excluded from participating in
valuable exchanges with others, e.g., participation in research teams with
leading researchers, access to the latest research findings, etc. This exclusion
can then negatively impact their career success (Tuire & Erno, 2001). As a
result, the production and sharing of valuable knowledge is facilitated, allowing
the frontier of knowledge to progress rapidly and at minimal cost (Liebeskind et
al., 1996).
thus the research also include researchers working for business firms, making it often hard to
distinguish between the two groups.
18
The majority of studies on scientific networks have investigated the physical sciences.
Garvey (1979) proposed that social science networks are similar to physical science networks,
although they may be less highly structured.
Studies of research-based communities of academic scientists have shown that

they are generally characterized by a center and a periphery (Schott, 1988)19.
The most important, visible, or active members are generally found in the
center, and these individuals influence the direction of the development of the
community’s knowledge. The activities of the individuals in the core determine
the community’s dominating theoretical concepts, methods, and chosen research
problems and these are then mediated through the network’s links to individuals
in the periphery (Schott, 1988). Through a process known as or social contagion
(Levy & Nail, 1993; Marsden, 1998), new members are socialized into the
community and as such transform their personal identities, adapting their
attitudes, behaviors, and values to those of the community (Holzner & Marx,
1979). Additionally, power is an integral part of scientific communities, with
individuals often using knowledge strategies as components of power strategies
(ibid). Thus, the center of a scientific community is not only a realm of activity,
but it also is a realm of identity and cultural values of the community (Schott,
1988; Tuire & Erno, 2001).
Work within the occupational community literature argues that individuals who
attain a position of centrality in the inter-organizational network generally have
a higher degree of knowledge, power, prestige, and honor (van Maanen &
Barley, 1984). Individuals who are visible in central positions are seen to be the
sages of the community. Similar to Wenger’s degrees of participation, within an
occupational community, newcomers have a time of apprenticeship or learning
in which they learn the “rules of the game”. It is suggested that centrality, work
performance, and individual careers are closely interconnected factors within
these networks. An individual’s reputation is gained through expertise in one’s
work, where one conducts his or her work – “majors” vs. “minor leagues”, and
whom one knows – e.g., doctoral student and faculty supervisor (van Maanen &
Barley, 1984). Thus, these professionally based inter-organizational networks of
practice are similar to communities of practice in that they are composed of
circles of members who have passed through various “boundaries”.
In addition to the above research that takes the perspective of the inter-
organizational network, researchers investigating participation in inter-
organizational networks of practice have also taken the firm’s perspective20.
This research has primarily focused on high technology industries and one area
of investigation is why individuals communicate informally with others outside
19
See Schott (1988) for a discussion of the center-periphery model.
20
Research has found that academic and industrial scientists display similar work-related
communication patterns. For a study that directly compares these groups within three
different fields, see Debackere & Rappa (1994).
44 CHAPTER TWO
the organization. First, von Hippel (1987) writes that when specialist engineers
cannot find the required know-how in-house or in publications, they go outside
their organization to their professional networks that are developed at
conferences and other events when individuals have the opportunity to judge
each other’s expertise. Further research has found that quite often professionals
communicate with others in their professional networks in order to maintain
contact with a professional reference group and to keep abreast of technological
changes (Aiken & Hage, 1968; Aldrich & Herker, 1977). However, Allen
(1970) also finds that low performing individuals choose to go outside for help.
He argues that this choice is a way to avoid paying a psychological price of loss
of face that occurs when an individual asks a colleague who is not a friend for
advice.
Extensions of this research within high technology-based industries provide

evidence that participation by individuals in inter-organizational networks of
practice leads to knowledge sharing across a firm’s legal boundaries and that
these activities are generally not governed by contracts or other market
mechanisms (Liebeskind et al., 1996)21. One of the most well known examples
of this is the study of Route 128 and Silicon Valley by AnnaLee Saxenian. In
this study, Saxenian proposes that one of the primary reasons for the relative
success of the Silicon Valley area is that knowledge is easily shared through
informal relationships between individuals belonging to competing firms. This
is in direct contrast to the Route 128 area in Boston where informal inter-
organizational fraternization is discouraged. These informal relationships often
form more rapidly than formal inter-organizational relationships, thus
facilitating the flow of knowledge between organizations (Brown & Duguid,
2000).
Of interest to management is that this flow of knowledge into the firm through
participation in inter-organizational networks of practice by firm members is
generally not a one-way street. Rather, individuals are often likely to exchange
21
Considerable research has also been performed on these inter-organizational networks but
from the firm’s point of view, e.g., inter-organizational boundary spanning activity - a major
stream began in the 1960s with the investigation into the communication patterns of scientists
and engineers in R&D laboratories (see e.g., Allen (1977), Allen et al. (1979) etc.; see Flap,
Bulder, & Völker (1998) for a review) and informal know-how trading – a relatively less
investigated stream of research, e.g., semiconductor, specialty steel and mini-mill industry,
and R&D operations (Schrader, 1991; von Hippel, 1987; Bouty, 2000). We discuss these
later in various sections of this thesis.
or trade knowledge22 with others who might even be working in rival firms
(Czepiel, 1975; David & Cochran, 1987; von Hippel, 1987; Schrader, 1991),
thus it is argued that knowledge “leaks” across the firm’s legal boundaries
(Mansfield, 1985; von Hippel, 1988). However, this research suggests that
individuals do not just give the knowledge away to others in their networks of
practice. Rather they consciously exchange knowledge with other carefully
chosen individuals with whom they often have a long-term relation built on
mutual trust and understanding (Schrader, 1991; Bouty, 2000). Research
conducted by Schrader (1991) finds that individuals often expect that their
chances of receiving valuable knowledge in return from the knowledge seeker
are likely to increase after they provide knowledge. Thus, participation in inter-
organizational networks of practice results in a feeling of reciprocity and a
dyadic exchange of knowledge (von Hippel, 1987; Macdonald & Williams,
1993) with knowledge sharing viewed as an ‘admission ticket’ to the ongoing
‘back room’ discussions within professional networks (Appleyard, 1996).
As a result, participation in inter-organizational networks of practice leads to

knowledge leaking out at the same time as it leaks in (Brown & Duguid, 2000).
However, research on the relationship between this knowledge trading and
performance at any level is scant. One of the primary reasons is that it is very
difficult for firms to manage and evaluate the benefits since it occurs “off the
books” (Carter, 1989). Secondly, data regarding the sharing of potentially firm
proprietary knowledge are difficult to collect due to their sensitive nature.
However, there is some initial evidence of a positive relationship between
knowledge trading and firm performance (Allen, Hyman, & Pinckney, 1983;
Schrader, 1991).
While science-based inter-organizational networks of practice have received

considerable attention from researchers, in recent years there has been an
increased interest in inter-organizational groups that are not necessarily science-
based or profession-based. Mentioned above, occupational communities may
also be inter-organizational (Van Maanen & Barley, 1984). However, we argue
that occupational communities are not the same as inter-organizational
distributed networks of practice since occupational communities comprise
individuals who belong to the community merely due to their sharing the same
occupation and not based on their emergent relationships based upon interaction
while performing work tasks. Thus, while an inter-organizational distributed
22
These studies often used the term know-how trading. In order to keep in line with the
terminology in this thesis, knowledge exchange will be used here to indicate know-how
trading.
46 CHAPTER TWO
network of practice may comprise individuals from the same occupation, not all
occupational communities are inter-organizational distributed networks of
practice. In addition, the literature on occupational communities has generally
seen these groupings in a negative light in relation to organizations in which
there is a constant power struggle between management and occupational
communities. One of the primary goals and differentiators of occupational
communities is argued to be self-control, or the occupational community’s
ability to dictate who will and will not be a member, how the content and
conduct of a member’s work will be assessed, as well as the bargaining power of
its members over management within organizations. For example, the literature
discusses unionization and professionalization as a means to gain and retain self-
control so that the collective career of the community may be advanced. When
knowledge is discussed, it is described as a means merely for the community of
generating and maintaining self-control and thus, is not the primary focus of the
group, in direct contrast to that of inter-organizational networks of practice.
Finally, of interest is that there have been very few studies investigating
occupational communities (Barley, 1996)23.
A second development in the area of inter-organizational networks worth

mentioning is that of epistemic communities. In its original definition, the term
was used to describe groups of individuals, such as those within the same
profession or scientific discipline, who share a common frame of reference and
set of perspectives with which they construct their reality (Holzner & Marx,
1979). This term has been adapted within the field of international relations to
describe groups of individuals who are bound together by “their shared belief or
faith in the verity and the applicability of particular forms of knowledge or
specific truths” (1992:3) and as such resembles Fleck’s “thought collectives” or
Kuhn’s paradigms. Within international relations, epistemic communities have
been used as a lens to understand policy changes and coordination at the
23
In one of the few studies on occupational communities, Lawrence (1998) conducted a study
of a small Canadian Forensic accounting community in a western Canadian province
(comprising 10 to 14 established accountants). This limited study of 22 interviews was based
on a sampling design that has been labeled as snowballing in which initial interviewees
identified successive interviewees and so on. Interviewees consisted of accountants, lawyers,
and law enforcement officers. Relevant results revealed that there were both formal and
informal rules for membership. The formal rules, such as membership in professional
associations, were considered a prerequisite for all potential members; however, the informal
rules were those that determined in essence who became an insider of the community. These
informal rules involved stereotypes that cast some people as insiders and others as outsiders
based on particular characteristics (e.g., sex, age, education, experience), personality, or social
processes (e.g., recommendation through word-of-mouth) that worked to include some while
excluding others.
national and international levels that have been impacted by communities of

specialists. These epistemic communities consist of individuals from a variety
of disciplines and backgrounds, have a shared commitment to enhance a
particular set of knowledge, and are motivated by the possibility to influence
public decision making in their field of expertise24. However, this set of inter-
organizational networks falls outside the scope of this thesis since they generally
comprise individuals such as governmental policymakers and as such are
primarily not working for business firms.
Beyond this, little research on inter-organizational networks of practice of

individuals who are not involved in science-based industries has been
conducted, thus leaving us with a limited understanding of these networks. One
of the reasons for this lack of research may be due to the relative difficulty in
identifying and studying these emergent networks due to a lack of public record
of communications or certain ethical issues in conducting research. In
comparison, scientific communities have been easier to research due to the
accessibility of publicly available data such as citations and bibliographic
references. However, with the global spread of the internet, electronic
communities have developed that provide a virtual space for individuals
regardless of organizational affiliation, profession, or status to communicate on
work-related matters. In addition, these electronic networks often conduct their
activity in the public domain on the internet, thus facilitating the ability to
research them. We turn to these electronic networks in the next section after we
summarize inter-organizational networks of practice.
In summary, the inter-organizational networks of practice described here are

similar in many ways to communities of practice: a smaller set of individuals
bound together by a common practice and acting communally, a focus on
knowledge and learning, strong social norms of behavior and control, and
relationships more of a dyadic reciprocal nature. However, there are some
differences. First, similar to the discussion above regarding organizational
locale as a differentiating factor between intra-organizational networks of
24
Haas (1992:3) proposes that members of epistemic communities have: “(1) a shared set of
normative and principled beliefs, which provide a value based rationale for the social action
of community members; (2) shared casual beliefs, which are derived from their analysis of
practices leading or contributing to a central set of problems in their domain and which then
serve as the basis for elucidating the multiple linkages between possible policy actions and
desired outcomes; (3) shared notions of validity – that is, intersubjective, internally defined
criteria for weighing and validating knowledge in the domain of their expertise; and (4) a
common policy enterprise -- that is a set of common practices associated with a set of
problems to which their professional competence is directed, presumably out of the
conviction that human welfare will be enhanced as a consequence.
48 CHAPTER TWO
practice and communities of practice, we find the same applies to inter-

organizational communities, yet to a higher degree. In inter-organizational
communities, individuals come from a variety of organizations, thus they tend to
be even more heterogeneous in their experiences. Different organizational
affiliations may also lead to a more weakly shared identity, language, norms,
and values (Fiol, 1991). A second difference between these two types of
networks focuses on the knowledge created. In scientific research–based
communities, knowledge created is more universal since it is less situated in a
physical location or organizational context. However, knowledge created within
a community of practice is more situated in a physical location and as such tends
to be less explicit and more difficult to diffuse for reasons described above.
Additionally, members of a community of practice focus on improving their
own competencies and practice and not that of other individuals who may be
working on similar tasks in other locations. Knowledge spillover for others
outside the community of practice is a by-product of communities of practice
and not the main focus. This is in contrast to the scientific and occupational
communities discussed above whose members generally work to improve both
their own competence as well as create knowledge for the greater set of
individuals involved in the same scientific specialty or profession.
Additionally, while we may know quite a bit about science-based and

profession-based inter-organizational networks of practice, we know little about
inter-organizational networks of practice comprising individuals outside these
groups. In other inter-organizational networks of practice, say commercial real
estate managers in Stockholm, we would expect to find differences. For
example, knowledge could be considered to be more of a private good since
there may not be channels or norms to make knowledge public. These
differences may then affect knowledge sharing within these networks.
2.4 Electronic Networks of Practice
As mentioned above, recent advances in internet communication technologies

have also led to the development of new forms of communication: emergent
electronic or virtual networks25. These electronic networks are the source of a
25
It is important to state our position on which perspective we take with regard to the use of
computer-based systems in social settings. In an extensive review of empirical studies
regarding computing use in organizations and public life, Kling (1980) contrasts two broad
perspectives: (1) systems rationalism and (2) segmented-institutionalism. Systems
rationalism includes a collection of approaches such as management science, management
rationalism, and the systems approach, and this perspective is found to be more helpful in
stable settings when there is consensus over important social values. Followers of this
high level of inter-organizational communication, such as listservs26, multi-user

domains27, chat rooms, and bulletin boards (Hinds & Kiesler, 1995; Constant,
Sproull, & Kiesler, 1996). For example, by 1999 the number of Usenet groups
expanded to more than twenty-five thousand different interest groups with more
than half a million postings per day (Dern, 1999). Firms are also leveraging
these new technologies and are implementing applications such as electronic
discussion networks or bulletin boards to promote knowledge sharing between
unacquainted individuals across the firm’s internal organizational boundaries
(Fulk & DeSanctis, 1995; Davenport & Prusak, 1998; Alavi & Leidner, 1999).
Similar to the previously discussed networks of practice, electronic networks of
practice are also typically emergent networks, forming around a specific
practice, such as criminal law, and they have been given a number of names:
webs of knowledge, electronic communities, online communities, electronic
networks, and even communities of practice28. In our discussion, we will use
some of these terms interchangeably to refer to electronic networks of practice.
Due to their relative recent appearance, there is a very limited amount of

research on electronic communities. Wellman & Gulia (1999) discuss the dearth
perspective view efficiency, whether economic or organizational, as the predominant value.

However, segmented-institutionalism is found to be more helpful in more dynamic settings
involving a wider variety of groups in which there is generally more conflict instead of
consensus on social values. Followers of this perspective view the predominant values to be
sovereignty of individuals and groups over the critical aspects of their lives, the integrity of
individuals, and social equity. Economic and organizational efficiency is subservient to these
values. Due to the dynamic aspects of electronic networks of practice and our argument that
individuals can make their own choices regarding the sharing of knowledge and degree of
participation in electronic networks of practice based on individual motivations of self-
interest and collective interest, we identify more with the segmented-institutionalists than with
the systems rationalists.
26
Listservs are mailing lists that forward email messages to everyone who has subscribed to
the list. Members seldom know who each other are due often to the large size of these lists.
27
Multiuser domains (MUDs) attempt to model physical spaces and face-to-face interaction
using text-based virtual realities.
28
Electronic communities can revolve around numerous topics and are not necessarily work
related. For a discussion of various aspects of electronic communities, see Smith & Kollock
(1999) or Lueg & Fisher (2003). In addition to work-related communities, there are four
main types of online communities (Hagel & Armstrong, 1997): (1) communities of interest:
individuals who share an interest, expertise, and passion in an area, e.g., bird watching, sport
scars, etc., (2) communities of relationships: individuals who need to share a personal life
experience, e.g., breast cancer, divorce, etc., (3) communities of transaction: individuals who
exchange information to facilitate economic exchanges, e.g., wine, and (4) communities of
fantasy: individuals who explore new identities in imaginary worlds of fantasy. As
mentioned, keeping in line with the focus of this thesis, we will focus our discussion only on
electronic networks of practice, or electronic communities that revolve around a specific work
practice.
50 CHAPTER TWO
of studies on online communities. As such, we have chosen to discuss intra-

organizational and inter-organizational virtual communities together in this
section, indicating differences where appropriate. We begin with a general
overview of electronic networks of practice.
The unique characteristic of electronic networks of practice is that they enable

the creation of weak structural links between thousands of geographically
dispersed individuals. Individuals may quickly and effortlessly access a broad
source of expertise through a wide variety of knowledgeable individuals with
whom they are not acquainted regardless of their demographic characteristics,
organizational setting, or local culture (Hinds & Kiesler, 1995; (Sproull & Faraj,
1995, Faraj & Wasko, 1998). In these networks, individuals engage in
knowledge sharing, problem solving, and learning through posting and
responding to questions on professional advice, storytelling of personal
experiences, and debate on issues relevant to the network (Wasko & Faraj,
2000). Individuals benefit from these networks since they gain access to and
even randomly come across new information, expertise, and ideas that are often
not available locally due to the extensive reach of these networks (Brent, 1994).
In addition, these networks enable individuals to gain access to the latest
thinking within their field.
In electronic networks of practice, anyone interested in the group’s topic may

join and the number of participants is unlimited. Little is known about the
participants other than an email address and what each individual voluntarily
chooses to disclose (see Sproull & Faraj, 1995; Kollock & Smith, 1996; Wasko
& Faraj, 2000). Examples of Usenet communities include those revolving
around more technical issues such as comp.lang.c++, comp.objects, and
comp.database. These technical electronic networks display characteristics
similar to scientific communities since they are open to anyone with the express
purpose of developing valuable programming knowledge in rapidly changing
technical fields (Wasko & Faraj, 2000). Participation in electronic networks of
practice is argued to be especially rapid in organizations with individuals who
are interested in maintaining interpersonal social ties or as described above,
weak structural links, that are based primarily on similar professional interests
(Pickering & King, 1995).
Comparing electronic networks of practice to communities of practice, they are

similar in that they are a social space where individuals working on similar tasks
self-organize to help each other and share perspectives about their occupational
practice or common interest (Brown & Duguid, 2000). However, they exist
primarily online and have a natural boundary based on membership, and we
believe that there are other critical distinctions between the characteristics of a
community of practice and an electronic network of practice. First,
communities of practice are generally characterized by rich face-to-face
exchanges in person-to-person interactions. As discussed above, the tightly knit
social structures of communities of practice facilitate the creation of a shared
identity, common language and norms as well as trust, obligation, and social
controls. These characteristics have been argued as essential for the continuous
incremental improvements in the community’s practice (Brown & Duguid,
1991, 1998; Wenger, 1998). However, interactions in an electronic network of
practice are limited to text-based, asynchronous, computer-mediated
communication. Theories of media richness argue that text-based computer-
mediated communication is a lean medium of exchange. In other words, it is
difficult for people communicating to develop a shared meaning and
understanding since this type of communication is impersonal, does not provide
immediate feedback, and does not provide additional meanings through body
language, voice inflections, dress, posture, or tone of voice (Sproull & Kiesler,
1991; Lee, 1994). Since members are not physically in the presence of each
other, norms are not as dominating in electronic communities, allowing for more
individual freedom in action (Squire & Johnson, 2000). Additionally,
individuals may be scattered geographically across the organization or across
numerous organizations and as such may not interact on a daily basis in a face-
to-face manner. This type of interaction leads to relatively little direct
reciprocity between individuals and as a result, reach dominates direct
reciprocity in these networks. In this manner, electronic networks of practice
resemble loosely coupled systems (Weick, 1979). As a result, the ability of
members to develop a shared identity and repertoire through narration,
collaboration, and social construction is hampered.
However, as mentioned, electronic networks of practice have a greater reach

than traditional networks of practice, and as such do support the creation of
weak electronic “bridging ties” between an unlimited number of like-minded
individuals. Discussed above, electronic networks of practice are not limited by
size constraints, and membership is typically voluntary and open, unlike
communities of practice. Anyone with an interest in the shared practice can
participate in an electronic network of practice regardless of social status, racial
demographics or geographic location because the technology filters out the
social cues associated with face-to-face interactions. Additionally, logistical and
social costs to participate are lower than in distributed networks of practice
(Sproull & Faraj, 1997). As a result, electronic network of practice participants
generally do not have personal familiarity with one another nor do they need to
have this in order to seek out answers and advice from other members. Thus,
52 CHAPTER TWO
unlike the previously discussed networks of practice where people know each
other personally and form dyadic, interpersonal relationships, participants in
electronic networks of practice are typically strangers, personal information is
limited to what an individual wishes to voluntarily disclose, and individuals
form weak ties with the entire network instead of with a select few. In addition,
the open and fluid membership in an inter-organizational electronic network of
practice enables dynamic interactions regardless of local organizational rules
and hierarchies. Thus, an electronic network of practice increases the likelihood
of connecting knowledge seekers with other knowledgeable helpers regardless
of interpersonal social ties, potentially increasing access to greater resources
than are available in a local community or in one’s network of acquaintances.
In electronic networks of practice such as those supported by listserv or bulletin

board technologies, it has been proposed that knowledge be considered as a
public good (Kollock, 1999). In the formal language of collective action theory,
the network participants are the interest group, and the public good is the
continuous stream of knowledge produced and jointly held by the network’s
participants. In an electronic network of practice, the posting of messages is
open to anyone and once posted, messages are visible to everyone participating
in the network. Thus, anyone searching for advice can either post a new
question, or in some networks search the archived discussions to reuse
knowledge that has already been exchanged between other members. This
ability to make all interactions visible and reach everyone in an electronic
network of practice contrasts with the dyadic interactions in a community of
practice or a traditional network of practice. Due to this open nature, the
network’s knowledge is non-excludable. When one participant responds to a
posting, then all members may benefit from this knowledge, even though they
did not contribute to its production through either posting or responding.
One central research area in electronic networks of practice is why do people

spend their time helping others and sharing their knowledge with others who are
typically strangers. When an individual shares knowledge with others, this
results in the loss of unique value of the individual relative to the others (Thibaut
& Kelley, 1959) and thus benefits all others except the individual (Thorn &
Connolly, 1987). Therefore, it seems irrational that individuals voluntarily
contribute their time, effort and knowledge to help strangers in a network of
practice (Wasko & Faraj, 2002). Researchers investigating this paradox have
found a variety of factors motivating online knowledge sharing, such as
organizational commitment, norms of reciprocity, and enjoyment in solving
problems (Constant et al., 1996), trust (Ridings, Gefen, & Arinze, 2002), and an
interest in advancing the community, intrinsic rewards, and increased reputation
(Lakhani & von Hippel, 2000). In addition, recent work has focused on
applying a public good perspective to the study of electronic networks of
practice. Using this perspective, it is argued that knowledge sharing is
motivated by moral obligation and mutual interest as opposed to self-interest.
There are two primary areas of debate for electronic networks of practice: (1)
whether they can be designed and managed and (2) whether learning and
innovation can be supported by them. Regarding the first area, authors
commonly focus on the technical aspects of designing and managing electronic
networks of practice while failing to discuss the difficulties due to the various
social aspects of the community (Hara & Kling, 2002). As for the second area
of debate, the recent phenomenon of open source software development projects
provides support the idea that knowledge creation can occur through virtual
means only. In these projects, individuals from across the world and
organizations create knowledge completely online through the development of
software programs, e.g., Linux, Apache. The intriguing aspect of these
communities is that individuals “freely” and “voluntarily” collaborate to develop
software that they or their organizations need (von Hippel & von Krogh, 2003).
The result is that the innovation no longer is an individual task, but a joint effort
produced by the “community”. Recent work examining the open source
software phenomenon has proposed a compound model of innovation, the
private-collective model in which private interests are combined with collective
interests, as a means to better understand this form of collaborative behavior that
results in knowledge creation (ibid). While these open source projects are not
electronic networks of practice, there are often electronic networks of practice
associated with them. For example, project founders often set up mailing lists
for individuals using or developing the specific software code to seek help, to
provide information, or to provide new code for others to discuss and test (ibid),
and as we will see in the next chapter have been the focus of several studies.
In summary, electronic networks of practice are similar to traditional networks

of practice since they are a social space in which individuals working on similar
tasks may collaborate and share knowledge. However, there are some
considerable differences. Electronic networks of practice are characterized by
generalized reciprocity, a high degree of reach, and a low degree of social
constraint and shared identity. As a result, while knowledge may be shared in
these networks, it tends to be of a much more explicit nature, and potentially of
a more novel nature, than that of non-electronic networks of practice. Finally,
one considerable difference is that knowledge within electronic networks of
practice can be considered to be more of a public good since all interactions
between members are generally visible to all members.
54 CHAPTER TWO
2.5 Summary
While this discussion has revealed that there are considerable differences
between the various networks of practice, they do have one important aspect in
common. Since they are all emergent networks, these various networks are self-
organizing and autonomous. Thus, the continued vitality of any network,
regardless of type, is dependent upon the willingness of individuals to
participate and share knowledge with one another. In summary, this chapter has
provided us with a basic understanding of the various networks of practice. In
the next chapter, we will review the relevant empirical studies conducted to date,
while we return to a comparison of the various types of networks of practice in
the final chapter.
CHAPTER THREE
Review of Previous Empirical Studies of

THIS CHAPTER REVIEWS the set of empirical studies that are relevant
to our study on networks of practice in order to develop the research
purposes of this thesis. Notice the use of the phrase “relevant to our study
on networks of practice” as opposed to “about networks of practice”. To
date, the number of studies that focus explicitly on networks of practice is
quite limited, especially within certain areas. However, there are several
bodies of literature that contain studies relevant to networks of practice
(e.g., technology transfer) due to their focus on emergent networks whose
main activity is knowledge creation and sharing. In figure 3.1, we present
the network of practice matrix in which we provide an overview of the
selected bodies of literature from which we draw.
Any review of this degree of breadth is necessarily selective, thus it is

important to note a few limitations. First, we focus our review only on
studies that rely on empirical fieldwork and not on theoretical articles or
studies based on experiments or simulations. This line has been drawn
since we are interested in what we “know” about networks of practice as
opposed to what we “think we know” about them. In addition, as
mentioned above, studies will be primarily limited to those that investigate
networks of individuals within and across business firms that emerge as
individuals interact on work-related tasks, and thus we will not include
studies on networks that are of a more formal nature nor those that emerge
due to non-work related interactions, e.g., friendship, common interests,
etc. In the few areas of the matrix populated by a higher number of studies,
we have selected for the most part only the major academic works of a
more reputable quality. Finally, we have chosen not to include the more
practitioner-oriented studies since these tend to be less grounded in theory
56 CHAPTER THREE
and less academically rigorous while focusing more on “how to” in order to
enlighten practitioners29. In conducting such a review, it is inevitable that
justice will not be done to all research, and as a result, some studies will
receive too little or no attention.
Figure 3.1 Selected Literatures for Network of Practice Review

Intra-organizaitonal
Co-located
1.Communities
of Practice 0/ 0/
Non-co-located
2.Social Networks
0/ Multinationals
R&D
Technology Transfer
4. Online
Communities
3.Scientific Communities
organizational
Invisible Colleges
Inter-
0/ R&D/
Technology Transfer
Know-how Trading
4. Online
Communities
Under each type of network of practice, we present the studies in

chronological order based on the date they were published. As mentioned
previously, since we are primarily interested in issues of structure and
performance, we will pay particular attention to these issues in this review.
Thus, at the end of the review of each type of network of practice, we will
summarize and discuss issues relating to structure and performance as well
as the findings relating to the cognitive aspects within each particular type
29
For discussions on how to establish and build networks of practice, see McDermott
(1999a, 1999b), Dixon (2000), Wenger (2000), Wenger et al. (2002), Ackerman et al.
(2003). An extensive practitioner-oriented study was conducted by the American
Productivity & Quality Center (APQC 2001) on how to build and sustain “communities
of practice”. See Kim (2000) for a practical discussion of how to build online
communities.
REVIEW OF PREVIOUS STUDIES 57
of network of practice30. A table is then provided with the studies

organized in chronological order under the three research foci of structure,
performance, and cognitive aspects, with findings specifically related to
structure and performance highlighted. Finally, at the end of this chapter,
we will synthesize this literature review and discuss the research findings.
The next chapter, Development of Research Purposes, will then build on
the results of this review of empirical studies.
3.1 Studies of Communities of Practice
We begin our review with a look at the literature on communities of

practice. As mentioned previously, these studies tend to be ethnographies
and case studies of role-based communities in which members have similar
jobs or occupations and are co-located within one physical site within the
same organization. In addition, some of these do not explicitly label the
groups observed as communities of practice since they were conducted
prior to the widespread acceptance of the term.
One of the first studies that laid the foundation for the development of
communities of practice was conducted by Lucy Suchman (1983). In her
very exploratory study performed in, 1979, Suchman conducted
ethnographic research of two office workers in the accounting office of a
large US corporation, paying attention to the actual work performed to
accomplish work tasks. Her study began from the observation that the
“specification of even the most routine clerical work as a schema of
procedures is an unsolved problem in automated systems design”. In her
study, Suchman found that the practical action performed by the two
workers to actually get their tasks done varied from the procedural
specifications for conducting the work. Thus, the primary conclusion of
her study is that the “smooth flow” of office procedures is not the work
itself, rather it is an outcome around which practitioners orient their work.
Julian Orr (1990, 1996) followed in Suchman’s footsteps by also viewing

service work as situated practice, and as mentioned above, Orr’s study
30
While there are many ways to categorize these studies, we have chosen these three
since they include the two main aspects in which we are interested: structure and
performance. Additionally, we have titled the third category, cognitive aspects, since
this category primarily relates to aspects other than structure and performance, such as
norms, symbols, identity, and values.
58 CHAPTER THREE
provides a “thick” (see Geertz, 1973) comprehensive description of the way

that individuals actually conduct their work31. For his doctoral studies, Orr
conducted an ethnographic study of Xerox’s service repair technicians by
following six technicians of the Silicon Valley District Office on their daily
rounds for twelve days. Although Xerox runs an extensive training
program and produces a considerable amount of documentation to help
technicians conduct their work repairing machines, Orr observed that
technicians did not find these formal means to be helpful. Rather, there
was “clearly a disparity between the tasks which they are told to
accomplish and the means which are said to be adequate to the task. The
technicians chose to give the task priority over means to resolve the
problems in the field any way they could, apparently believing that
management really wanted accomplishment more than strict observation of
the prescriptions of work” (1990:15-16). Orr found that war stories, or
anecdotes of experiences in which problems and their solutions are
described using context and technical detail, are a prominent feature used
by the technicians when conducting their work. Additionally, Orr observed
that informal meetings such as the breakfasts that technicians attended were
valuable fora for discussing work and narrating their war stories.
Lave & Wenger (1991) took somewhat of a different approach than the
above two studies and focused their study on the relationship between
learning and social situations, and in so doing developed the concept of
situated learning. In their frequently cited study that was primarily based
on the research of others, they investigated apprenticeship by Liberian
tailors (Lave, 1988), Mayan midwives (Jordan, 1989), non-drinking
alcoholics (Cain n.d.), butchers in US supermarkets (Marshall, 1972), and
U.S. navy quartermasters (Hutchins, 1996). Based on these studies, Lave
& Wenger contrasted the traditional dyadic view of learning of a student
and teacher with the concept of legitimate peripheral participation within
communities of practice. Lave & Wenger describe how apprentices (or
newcomers) participate peripherally in a practice, learning from the masters
(old-timers) and more experienced individuals (young masters or
journeymen) within a community of practice. Thus, structure for Lave &
Wenger is more an adaptive outcome of action rather than a precondition
within a social system. Touching on structure, they argue that a
community of practice has no place designated the “periphery” nor a single
31
See Burawoy (1979) for an historical overview and interesting example of
anthropology in the workplace.
core or center. In terms of performance, Lave & Wenger propose that full
participants of a community of practice are those who have acquired
legitimacy as well as the skills, knowledge, and discourse required by the
practice. In addition, the power dynamics among individuals in
communities of practice are discussed.
Referred to by Lave & Wenger above, Hutchins (1991, 1995, 1996)

studied socially distributed cognition in an in-depth, impressive study of
the work practice of a formally structured navigation team on a U.S.
warship. In the above work, Lave & Wenger (1991) view this team as a
community of practice. This is exemplified when the team is issued a new
officer and the community of practice provides the informal forum for
learning since one of the petty officers who is lower in rank yet has more
experience supervises the higher ranking officer. As such, this work
illustrates how a formal team can develop into a community of practice
over time as the team members interact while conducting their tasks.
In another study on distributed cognition, Scott Cook & Dvora Yanow

(Cook & Yanow, 1993; Yanow, 2000) performed a set of extensive case
studies investigating organizational learning32 at three small flute
companies near Boston, the most well known being the Powell Flute
Company. Each of these companies comprised about 25 individuals, with
only one or two individuals not directly involved in flute production. In
their study focused on distributed cognition, the important observation is
that tacit knowledge is shared and held without being made explicit at the
collective level through interactions revolving around the flutes by the
group members. This observation is in direct contrast to previous work by
Argyris & Schön (1978) who proposed that cognitive maps should be made
explicit in order for learning to take place. In revisiting the flute study,
Yanow (2000) further argues that some tacit knowledge should not be
made explicit and that doing so could at times even be harmful.
One of the studies to first explicitly discuss communities of practice and in

an attempt to link them with performance, Bill Snyder (1996) wrote his
32
The literature on organizational learning discusses situated learning, and it has
recently been argued by some that communities of practice are the appropriate unit of
analysis for organizational learning. For discussions and reviews of organizational
learning, see Huber (1991), Easterby-Smith, Araujo, & Burgoyne (1998), and Crossan
& Guatto (1996).
60 CHAPTER THREE
doctoral thesis on the relationships between organizational learning and

organizational performance. He conducted a rigorous case study of two
divisions at four regional offices of the US Veterans Benefits
Administration (VBA), with offices employing between 100 and 250
individuals. In his study of VBA office workers ranging from staff
members and clerks to supervisors, he observed communities of practice of
varying strength and effectiveness within the offices as well as intra-
organizational distributed networks of practice spanning the offices.
Snyder observed that the capacity of office workers to share knowledge and
skills was reduced when they belonged to a community of practice whose
members did not feel a high level of shared trust and commitment.
Looking at the degree to which teams met customer service objectives,
Snyder argues that communities of practice are critical to organizational
learning activities and as such influence organizational performance
through their impact on the creation and sharing of organizational
knowledge. Thus, Snyder’s results indicate that the levels of trust, respect,
and mutual commitment as evidenced in the strength of the relationships
among members of a community of practice affect the sharing of
knowledge and skills among community of practice members and thus
indirectly organizational performance.
In the most extensive study devoted to communities of practice, Wenger

(1998) observed the learning processes regarding the tasks of filling out
claims forms of around 20 co-located insurance claims processors in a US
firm. As discussed in Chapter Two, Wenger thoroughly develops the
definition of communities of practice, with a focus on discussing the
various related cognitive concepts such as practice, meaning, community,
learning, identity, and participation. With regard to structure, Wenger
elaborates on Lave & Wenger’s discussion, stating, “a community of
practice is a node of mutual engagement that becomes progressively looser
at the periphery, with layers going from core membership to extreme
peripherality” (1998: 118). In terms of performance, in Wenger’s view,
individuals in the community of practice’s core are those who have
mastered the tasks of claims processing while those in the periphery
perform the tasks to a lesser degree the further from the core the individual
is. However, beyond this, Wenger offers little discussion on these two
areas.
In response to the above work focusing on only a single community of

practice at a time, studies have begun to appear recently that focus on the
sharing of knowledge between communities of practice. Building on work
by Star & Griesemer (1989) on boundary objects, these studies find that the
use of language and boundary objects in the form of codified objects, such
as artifacts and texts, facilitate the sharing of knowledge that is situated in
local practice between communities of practice. In an impressive doctoral
study that resulted in an Organization Science publication, Carlile (1997,
2002) conducted an ethnographic investigation of a small co-located firm
(about 300 people) in the US auto supply industry. Carlile focused on
understanding how work practice shapes knowledge and the boundaries
between practices as well as on identifying the activities or processes that
are effective in facilitating collaboration across these boundaries. Based on
his observations of collaboration between the sales, design engineering,
manufacturing engineering, and production departments, Carlile discusses a
pragmatic approach to understanding knowledge sharing processes,
proposing that knowledge is localized, embedded, and invested in practice.
He observes that knowledge is both a barrier as well as a source of
innovation in a product development setting and similar to the next study,
he found that the use of boundary objects helped resolve the problems
arising when collaborating across community boundaries.
In a second study investigating knowledge sharing, Bechky (1999, 2003)

performed an award-winning ethnographic study33 of a US high-technology
manufacturing firm for her doctoral studies. In her investigation of
knowledge sharing between three co-located functional communities:
engineers, technicians, and assemblers, she found that each community had
a different understanding of the machines that they were building as well as
different languages regarding the machines and their tasks. As a result,
misunderstandings in production occurred between these communities.
Her findings suggest that knowledge sharing and learning between
communities is dependent upon the presence of individuals who speak the
various languages of the communities in question and upon individuals
keeping communication focused on the concrete. Her study thus indicates
that the knowledge of communities of practice can be captured,
33
1999 William Newman best dissertation paper award, Academy of Management,
1999 Louis Pondy best dissertation paper award, Organization and Management Theory
Division, Academy of Management.
62 CHAPTER THREE
disseminated, and preserved by organizations through the use of boundary

objects and brokers.
In yet another doctoral dissertation, Hara (2000) conducted a unique

ethnographic case study comparing several networks of practice (electronic
and face-to-face). Using data from observations, interviews, and document
reviews, Hara investigated the sharing and construction of knowledge by
lawyers in two public defender offices in Indiana and within an electronic
network of practice for all public defenders in the state. The primary
contribution of this work is in the area of a community of practice’s
knowledge. Hara critiques Lave & Wenger’s study of apprenticeship by
arguing that a community of practice’s knowledge is not static, mastered by
full participants and only to be learned by newcomers to the community.
Rather, Hara suggests that a community’s knowledge continuously evolves
as members interact with one another and as they come and go. Thus, all
members, regardless of their participatory status, continuously engage in a
learning process. Additionally, she suggests that a community’s
knowledge encompasses cultural, practical, and book knowledge. One
final finding is that only those lawyers who were physically co-located
developed a common identity due to their ability to informally socialize
and interact. This was despite the efforts by the public defender’s offices
to facilitate interaction between non-co-located lawyers through several
electronic means, e.g., listserv and other computer conferencing tools. This
further supports the importance of face-to-face interactions in the
development of communities of practice. We return to Hara’s work in the
section below on electronic networks of practice.
Gherardi & Nicolini (2000, 2002) further investigated the concept of

legitimate peripheral participation in a small ethnographic study of how
safety is mastered by novices on an Italian building site. In this study with
a slight bend towards practitioners that provides little new insight, these
researchers discuss how novices learned the practice of safety through an
interactive process of conversing and learning the language that included
observing body language, looking and seeing, and doing. Thus, one
finding of this study is that for some practices, the ability of a novice to
learn the tacit knowledge of the community is dependent upon the novice’s
co-located physical presence with other community members.
While the above studies primarily focused on understanding the cognitive

processes of communities of practice, with the exception of Hutchins
(1995) mentioned above, few studies have investigated the impact of the
formal organization on the formation of communities of practice in
organizations. In another doctoral study, Schenkel (2002) conducted an
extensive case study of a major Scandinavian construction project
involving 137 participants in managerial and support functions. He found
that the disciplining elements of ISO 9000 constrained the formation of
communities of practice within the project’s ten departments ranging
between 10 to 30 people; however, they encouraged the formation of a
network of practice across the departments.
In the last study on communities of practice to be reviewed, Schenkel &

Teigland (2002) conducted an exploratory study on the same construction
project mentioned above. This study provides mixed evidence for a
positive relationship between communities of practice and organizational
performance as measured by learning curves.
3.1.1 Discussion of Community of Practice Studies

In summary and as can be seen in table 3.1, the studies to date have
provided us with an extensive definition of the cognitive side of
communities of practice. Initial studies leading to the development of
communities of practice focused on understanding work practices and the
social nature and situatedness of learning by individuals. As time
progressed, researchers turned their focus more towards the knowledge of a
community of practice and how knowledge is shared both within a
community of practice as well as between communities of practice. One
main observation is that these studies provide support for the importance of
physical co-presence in the development of a community of practice. The
development of communities of practice is dependent upon the ability of
individuals to spontaneously socialize in an informal manner or to
unconsciously observe and learn from one another. Thus, while
information technologies may facilitate interaction between individuals
who are not co-located, they do not provide the same opportunities as co-
location does. This inability hampers the development of a common
identity and language between individuals and thus the development of a
community of practice. A second observation is that the formal
organization may impact the boundaries of communities of practice.
64 CHAPTER THREE
Functional boundaries as well as formal procedures affect the ability of

individuals to interact with one another by creating boundaries that are
either physical or intangible, such as those based on task knowledge.
Finally, when knowledge is to be transferred across the boundaries of a
community of practice, problems arise due to the situatedness of the
knowledge in the local practice. However, the use of boundary objects and
brokers who speak the language of the communities in question are useful
means of facilitating this transfer.
While we have a developed a relatively thorough understanding of the

above, there are several areas on which the empirical studies have failed to
touch. For example, these studies do little to shed light on the important
issues of structure and performance. As can be seen in table 3.1, no studies
have specifically focused on structure, and only two have a primary
research focus on performance. However, one of these is very exploratory
(Schenkel & Teigland) and the other (Snyder) focuses on the indirect link
between communities of practice and organizational learning, and not
organizational performance. Thus, as for the claims of the general
community of practice literature that these networks are linked to
organizational learning and performance through incremental innovation in
practice or that innovation occurs at the interstices of communities of
practice, the empirical studies to date provide little support for these claims.
Furthermore, these studies neglect other important aspects such as

community of practice lifecycles, the relationship between communities of
practice and management or organizational strategy, and power
relationships between members. As mentioned above, Lave & Wenger
(1991) discuss power and power relations in their book, yet the majority of
the subsequent work on communities of practice including Wenger (1998)
has failed to discuss these aspects in depth. Fox (2000) highlights this
point and proposes that Foucault’s work and actor-network theory (ANT)
could complement the community of practice literature by bringing in
aspects of power and inequality.
Finally, while most of the studies above are academically rigorous, there
are some concerns regarding generalizability. As evidenced above, the
number of empirical studies on communities of practice is very limited
both in methodology (ethnographies and case studies) and samples
(primarily the US and lower level, non-professionals in non-science
professions). Thus, this raises the question of generalizability of

community of practice theory to business firms outside of the US or to
professionals.
Table 3.1 Selected Studies of Communities of Practice (CP) *

Primary
Research Study Methods Sample Primary Findings
Focus
Structure
-------------
Performance
Relationship Snyder Case study: US office Communities of
between 1996 observations, workers in two practice influence
organizational interviews, divisions of four organizational
learning and archival data offices in US performance through
organizational review Veterans impact on
performance Benefits organizational
Association (100 knowledge and
to 250 learning. Sharing of
employees in knowledge and skills
each office) related to levels of
commitment and trust
among CP members.
Relationship Schenkel Survey Major Limited support for
between & Teigland Scandinavian positive relationship
community of 2002 construction between CPs and
practice and project learning curve
organizational involving 137 improvement.
performance managers and
support
individuals.
Cognitive
Aspects
“Smooth Suchman Ethnography Two office “Smooth” office
flow” of office 1983 workers in procedures are
procedures accounting outcome of actual
office of a large work conducted by
US corporation office workers and not
reflection of enduring,
externally decided
procedures.
66 CHAPTER THREE
Primary
Focus
What is work? Orr 1990, Ethnography Six Xerox Clear disparity
1996 service between the work
technicians from formally defined by
Silicon Valley the employer and the
tasks performed by the
technicians to
complete their job.
War stories of
significant importance
as diagnostic tool.
Relationship Lave & Ethnography Small groups of Learning is a situated
between Wenger based Mayan activity. Newcomers
learning and 1991 primarily on midwives, participate in
social research by Liberian tailors, communities of
situations others US non-drinking practice and learn
alcoholics, through legitimate
butchers in US peripheral
supermarkets, participation (LPP),
and US navy mastering the
quartermasters knowledge and skills
of the community.
Structure is more an
adaptive outcome of
action rather than a
precondition within a
social system. High
performers of practice
are full participants of
community, having
acquired legitimacy as
well as the required
skills, knowledge, and
discourse.
How do Hutchins Case study Navigation team Cognition is socially
people know 1991, on US warship distributed. Formal
what they 1995, 1996 team developed into
know? community of practice
over time.
Nature of Yanow & Case studies Craftsmen in Tacit knowledge can
learning by a Cook three small US be held at collective
collective 1993, flute makers level and need not
Yanow (approx. 25 necessarily be made
2000 people in each explicit to be shared
firm) within community.
Relationship Wenger Ethnography Around twenty Learning is a social

between 1998 claims phenomenon with
learning and processors co- knowledge as
social located in one competence gained
situations US company through participation
in a practice through
which meaning is
experienced.
Structure of
community of practice
consists of core of full
participants
surrounded by layers
of peripheral
members. Core
members are high
performers of CP
tasks.
Transfer of Carlile Ethnography Four CPs (sales, Knowledge is situated
knowledge 1997, 2002 design in local practice, thus
across engineering, problematic when
functional manufacturing transferred across
boundaries engineering, and functional boundaries.
production) in Boundary objects
US auto supply facilitate process.
firm of 300
people in one
site
Transfer of Bechky Ethnography Three CPs Language barriers are
knowledge 1999, 2003 (assemblers, significant between
across engineers, and communities, thus
functional technicians) in difficulties in
boundaries one site of US transferring
high technology knowledge between
manufacturer them. Boundary
objects facilitate
process by making
problems concrete as
well as individuals
who speak languages
of different
communities.
68 CHAPTER THREE
Knowledge Hara 2000 Ethnography Lawyers in two Community of

sharing within Public practice knowledge is
a community Defender’s not static, thus all
of practice Offices in members continuously
Indiana learn through
interaction.
Learning of Gherardi & Ethnography Italian builders LPP involves physical
tacit Nicolini on one presence due to
knowledge 2000, 2002 construction site conversing and
learning the language
through looking,
seeing, and doing.
Management Schenkel Case study Major Formal organization
of unexpected 2002 involving Scandinavian can influence CP
deviations in interviews, construction structure. ISO 9000
work survey, and project constrained formation
procedures archival data involving 137 of CPs within
review managers and construction site but
support encouraged formation
individuals of NP across sites.
* Findings specific to structure and performance are highlighted in bold text.
3.2 Studies of Intra-organizational Distributed Networks of

Practice
As mentioned in the previous chapter, intra-organizational distributed

networks of practice are networks of emergent relationships of individuals
who are distributed across an organization yet who work on similar tasks
using a similar base of knowledge. They are networks of individuals who
are acquainted with one another, have generally met face-to-face, and
communicate through both a mix of communication channels from face-to-
face meetings to electronic means. Furthermore, they are to be
distinguished from formally mandated teams. Being somewhat more of a
challenge to study due to their emergent nature and difficulties in
identifying them, there are very few empirical studies of these groups34.
However, by broadening our scope in terms of bodies of literature, we find
that there are a few studies on intra-organizational networks within the
34
There are several studies for and by practitioners on how to support the creation of
intra-organizational networks of practice through various means as described in the
above. However, as mentioned we have limited this review to studies that are of an
academic nature. See Wenger et al. (2002) for examples of management supported and
maintained intra-organizational networks of practice.
innovation, social network, and multinational corporation literature that are

worth mentioning35.
The only study of which we are aware that specifically focuses on intra-
organizational networks of practice and that is not explicitly practitioner-
oriented is by Hildreth, Kimble, and Wright (2000). Unfortunately, this
study is of a less rigorous academic nature than the previous community of
practice studies. These researchers focused on two networks of practice:
actuaries and IT support managers within two UK-based multinationals
with operations in Europe and Japan. In these cases, they found support for
the need for periodic face-to-face meetings to maintain levels of trust and
shared identity among network of practice members. In addition,
individuals who were separated by greater time distances from the majority
of the other members of the network felt themselves to be on the physical
periphery of the network. They felt that this physical separation hampered
their ability to truly participate in the network. A third finding that echoes
the findings in the community of practice literature is that the use of
boundary objects in the form of shared documents facilitated the sharing of
knowledge between the members.
One practitioner study worth mentioning due to the considerable attention

it has received despite its being published in the IBM Systems Journal is
that by Lesser & Storck (2001). These researchers interviewed five to ten
members in each of seven communities of practice and intra-organizational
distributed networks of practice that were of a more formal nature in US
firms (urban services specialists, land and real estate specialists, quality
champions in a manufacturer company, research chemists in a
pharmaceutical firm, programmers in a software development company,
researchers in a specialty chemical company, and project managers in a
telecom company). With a focus on performance, they suggest that
networks of practice may lead to decreased learning curves for new
employees, quicker responses to customers, reduced rework and avoidance
of reinventing the wheel, as well as the development of new ideas for
products and services.
Within the social network literature, there are three studies worth noting.
Han (1996) conducted a study that looked at the impact of the formal
35
For reviews of intra-organizational networks, see Krackhardt & Brass (1994),
Galaskiewicz (1996), Flap et al. (1998).
70 CHAPTER THREE
organization on the creation of distributed emergent networks. This

detailed case study of 76 employees in a large US retail corporation with
four hierarchical levels and numerous divisions found that formal
organizational boundaries affected the building of intra-organizational
relationships. Relationships were found to occur within the divisions as
opposed to across divisions and in general, the higher the individual was in
the hierarchy, the more relationships the individual had across divisions. In
the second study worth noting and as mentioned previously, Friedkin
(1980) found that the strength of the tie between two individuals had a
positive relationship with the degree of overlap between the contact circles
of the two individuals. He conducted this case study on scientists in seven
biological science departments belonging to the same U.S. university using
a questionnaire.
The third social network study is by Lazega (2001) who conducted an

extensive case study of 71 lawyers in the three offices of a North-Eastern
US corporate law partnership. Using data on advice, goodwill (co-
workership), and friendship relationships, Lazega found that these
relationships resulted in the creation of social niches (similar to intra-
organizational networks of practice), or stable quasi-groups that offered
members resources at a low cost, a sense of identity and common long-term
interests, and the stimulation needed to work together productively.
Lazega analyzed in detail aspects of status and social control and found that
underperformers were brought back into line through pressure placed on
them by other niche members. Additionally, he observed that those
lawyers who were informally sought out for advice and for collaboration by
others across the firm earned more money for the firm; however, he found
no relationship between those who were popular in the friendship networks
and individual economic performance.
Within the innovation and technology transfer literature, a series of studies

on communication patterns between engineers and applied scientists within
R&D operations has been conducted. Reflecting on this research, these
studies are relevant to the research on networks of practice despite their
being conducted several decades ago. In essence, these studies investigated
networks of practice since they focused on knowledge flows through
emergent relationships by studying interactions based on technical or
scientific conversations related to work tasks36. These studies began in the

1950s with work by Allen and his doctoral student, Tushman (see e.g.,
Allen, 1977). Their work is quite impressive in terms of depth and
methodology and they are among the few in the networks of practice
literatures to incorporate social network techniques. While there are
numerous studies performed within this area37, we will only mention the
ones most relevant to the research at hand.
Allen’s (1977) research focused on the study of 29 R&D project teams, and
he conducted network studies in thirteen different laboratories (smallest
being 20 professionals, largest being 400 professionals, all US except one
European) while Tushman (e.g., 1977) conducted an extensive field study
of a physically isolated R&D facility of a large U.S. corporation (345
professionals among total of 735 employees and 60 projects in seven
divisions). These studies are quite impressive in terms of their data
collection methods, with all professionals asked to keep “personal contact
records” for one day a week for a number of weeks to create the
communication networks. Analysis of the data revealed three types of
individuals who were central in the information flows: (1) communication
stars: individuals who were most central in these informal communication
networks, i.e., were most frequently approached by other colleagues for
information, (2) boundary spanners: individuals who spanned both intra-
and inter-organizational boundaries, (3) gatekeepers: communication stars
who were also inter-organizational boundary spanners.
With regard to the characteristics of these individuals, a significant degree

of overlap between individuals who conducted communication star,
boundary spanning, and gatekeeper activities was found. Thus, individuals
who were communication stars were more likely to be boundary spanners
as well as gatekeepers. These individuals tended to occupy higher
positions in management levels or were higher technical performers, i.e.,
36
In addition, these works are grounded in the information processing perspective on
organizational design. This view argues that individuals are constantly presented with
problems accompanied by various environmental and task uncertainties, and in order to
deal with these uncertainties, individuals must acquire information. Thus, the
information processing literature is similar to that of the network of practice literature
since these literatures both view individuals as unable to solve the tasks at hand based
only on the knowledge that they have in their head. For a look at the information
processing perspective, see Galbraith (1973) and Nadler & Tushman (1988).
37
For a review, see Aloni (1985).
72 CHAPTER THREE
more patents, very active publishers, and longer experience in the

laboratory. Thus, these individuals were seen to be more technically
competent with a higher level of work-related expertise than less active
individuals (Allen & Cohen, 1969; Tushman & Scanlan, 1981).
Another significant finding in this research stream is the relationship

between the degree of participation in “intra-organizational distributed
networks of practice” by project members and project performance. These
researchers found that this relationship was contingent upon the type of
project task being performed38. For example, higher performing product
development projects whose tasks were more locally defined were more
likely to have external project communication dominated by internal
boundary spanners, or individuals who spanned intra-organizational
boundaries (Tushman & Katz, 1980). On the other hand, research projects,
whose tasks were more universally defined, performed better if all
members participated in boundary spanning. Thus, this literature argues
that when local knowledge is not sufficient to complete the tasks at hand
and is of a more local nature, the organizational hierarchy is bypassed by
gatekeepers who rise to fill the need (Allen, Tushman, & Lee, 1979).
This research has implications for the network of practice literature since it
indicates that there is a relationship between participation in distributed
networks of practice, performance, and task knowledge. However, one
drawback of this research is that researchers primarily looked only at the
access of knowledge in the networks and not the providing of knowledge
by boundary spanners and gatekeepers. In other words, they did not look at
knowledge exchange or reciprocal actions. In addition, in the majority of
the articles published in the more reputable journals, only ego-centric data
are used. In other words, the researchers did not look at the position of the
38
Tushman (1977) defined four types of tasks: (1) basic research: work of a general
nature intended to apply to a broad range of applications or to the development of new
knowledge about an area, (2) applied research: work involving basic knowledge for the
solution of a particular problem. The creation and evaluation of new concepts or
components but not development for operational use, (3) development: the combination
of existing feasible concepts, perhaps with new knowledge, to provide a distinctly new
product or process. The application of known facts and theory to solve a particular
problem through exploratory study, design, and testing of new components or systems,
and (4) technical service: cost/performance improvement to existing products, processes
or systems. Recombination, modification and testing of systems using existing
knowledge. Opening new markets for existing products.
project or the individual within the entire intra-organizational distributed

network of emergent relationships.
Within the multinational corporation literature, emergent relationships have

been found to be a significant coordination mechanism between
geographically dispersed divisions. In their review of the literature on
coordinating mechanisms in multinationals, Martinez & Jarillo (1989)
found that researchers have been paying increasing attention to the
importance of “informal communication”, defined as informal networks,
personal contacts, intra-community visits, meetings, conferences and
forums, and transfer of managers, as coordinating mechanisms. Martinez
& Jarillo observed that thirty of the eighty-five pieces of research published
in books and journals that they reviewed discuss informal communication
and that 1976 can be seen as a turning point when researchers enlarged
their focus on mechanisms to include more informal mechanisms.
However, as evident in the title of Martinez & Jarillo’s work, the focus of
these studies is on the integration of activities. Few focused on the sharing
of knowledge through these means nor were intra-organizational
distributed networks of practice explicitly the focus of the research.
Subsequent research has focused on knowledge sharing through these
networks within multinationals. However, again the focus has not
explicitly been on networks of practice. For example, Tsai & Ghoshal
(1998) and Tsai (2002) did not really investigate networks of practice in
their study of multinationals since the level of analysis is the multinational
unit and not the individual or the network. For example, in the
operationalization of these studies, only unit managers were surveyed about
their units’ knowledge sharing and social relations (e.g., socializing during
events such as company picnics) with other units.
Despite the operationalization within this set of studies, one impressive

study worth noting is that by Morten Hansen during his doctoral studies.
Hansen (1996) built on the work by Allen, Tushman, and others in the
technology transfer literature and conducted a study of 120 projects in the
R&D operations of a technology-intensive multinational. Using
sociometric methods and critiquing the work by Allen and his colleagues,
Hansen looked at the position of the project within the firm’s entire project
population. Interestingly and which may be questioned, Hansen surveyed
only the R&D division managers and project team managers about
participation in emergent relationships at the division level and not at the
74 CHAPTER THREE
individual level since he argued that the relevant relationships were held at
the division level and not at the individual level. Based on a sophisticated
analysis of this unit level network, Hansen found a significant relationship
between project performance and the position of the project team’s division
in divisional networks of practice. Project teams whose divisions had
weaker ties within these networks of practice were more likely to achieve
shorter completion times when the knowledge to be transferred was of a
less situated nature (more codified and less tacit). For teams with stronger
ties in the network, however, there was a net effect in terms of completion
time. Strong ties facilitated the sharing of more situated knowledge for
such teams, yet norms of reciprocity meant that the teams were then
expected to return the help, thus slowing their completion times.
3.2.1 Discussion of Intra-organizational Distributed Networks of

Practice
In summary, there are very few in-depth quality studies specifically
focused on intra-organizational distributed networks of practice, and the
limited research reviewed here touches only briefly on the areas of
structure and performance. The innovation, technology transfer, and
multinational literature provides suggestive evidence of a relationship
between participation by individuals in intra-organizational distributed
networks of practice and performance, yet research suggests that the formal
organization impacts the emergence of these relationships across an
organization’s physical locations. As can be seen in table 3.2, these studies
were either conducted twenty to thirty years ago, are of a questionable
academic rigor, and/or are not specifically focused on intra-organizational
distributed networks of practice. In addition, problems of generalizability
surface when we note that the more rigorous studies comprise primarily
scientists, researchers, and lawyers in US settings. Interestingly, there are
no studies specifically addressing the cognitive aspects of these networks.
Thus, despite such claims by researchers that individuals participating in
these networks are able to share knowledge and develop a shared identity,
values, and language, with positive effects for performance, we have little
substantive evidence for these claims.
Table 3.2 Selected Studies of Intra-organizational Distributed

Networks of Practice (IANP) *
Primary
Focus
Structure
Structure of Friedkin Sociometric Scientists in The stronger the tie
weak ties 1980 survey seven biological between two
departments in individuals, the
US university higher the overlap of
their contact circles.
Factors Han 1996 Sociometric 76 employees in Formal
affecting survey large US retail organizational
creation of corporation boundaries affect
emergent building of intra-
relationships organizational
relationships
Collective Lazega Case study 71 lawyers in Advice, goodwill (co-
action among 2001 involving three offices of workership), and
individuals sociometric US law friendship
equal in data partnership relationships create
power social niches, or stable
quasi-groups that offer
members resources at a
low cost, a sense of
identity, common
long-term interests,
and the stimulation
needed to work
together productively.
Social pressure is
applied by others
within social niches to
maintain individual
performance levels.
Individuals who most
sought out for
collaboration and
advice by others across
the firm earn more
money for the firm.
76 CHAPTER THREE
Primary
Focus
Performance
Relationship R&D Sociometric Primarily US Hierarchy bypassed
between studies, surveys R&D when local knowledge
information e.g., Allen, laboratories and not sufficient for tasks.
flows and Tushman, their projects Project performance
project etc. (1960s, dependent upon
performance 1970s) project task
knowledge and
number of
gatekeepers.
Individuals highly
involved in IANPs
also central in local
laboratories. These
individuals generally
in management
positions or highly
technically competent
and seen as experts.
Knowledge Hansen Sociometric 120 R&D The more central the
integration 1996, 1999 survey projects within R&D team in
across one US organizational
subunits in multinational network in terms of
multiunit firm team’s unit possessing
relevant expertise, the
easier the network
search, and the faster
the completion time.
Weak network
relations slow down
projects when
knowledge transferred
is very complex.
Relationship Lesser & Interviews Five to ten Support for positive

between Storck members of relationship between
networks of 2001 seven IANPs participation in
practice and comprising IANPs and
creation of knowledge organizational
organizational workers in a performance.
value variety of large Participation in IANPs
corporations results in decreased
(majority likely learning curves for
in US) new employees,
quicker responses to
customers, reduced
rework and avoidance
of reinventing the
wheel, development of
new product and
service ideas.
Cognitive
Aspects
-----------
3.3 Studies of Inter-organizational Distributed Networks of

Practice
Turning now to inter-organizational distributed networks of practice, we

find several bodies of relevant empirical studies such as those on scientific
communities. While there is a wide body of research on formal inter-
organizational networks as well as formal inter-organizational boundary
spanning activity, as noted above, these studies will not be included in this
review due to their formal element39. We start our review with some of the
relevant studies from the scientific communities before taking a look at
some of the major studies performed from the viewpoint of the firm.
3.3.1 Scientific Community Perspective

Until the work by Diana Crane (1972), much of the work investigating
scientific communities used bibliographic methods such as references and
39
For a review of inter-organizational networks, see Mizruchi & Galaskiewicz (1994).
The literature on formal inter-organizational boundary spanning has its roots in the
work by Katz and Kahn (1966) and Kahn et al. (1964) and for reviews of this research,
see Van de Ven (1976), Aldrich & Whetten (1981), Galaskiewicz (1985), and Oliver
(1990).
78 CHAPTER THREE
citations, with co-citations being one of the most commonly used40.

However, the use of co-citations is no guarantee that these individuals are
interacting with one another. Garfield, Malin, & Small (1978:186) state,
“None of the bibliometric linkages [including co-citation analysis] require
that social contacts lie behind them, but the existence of strong patterns of
coupled documents (clusters) suggests that underlying social factors are at
work.” Thus, this review covers only the more important works that
actually investigate scientific communities based on interpersonal
interactions and not those that focus solely on bibliometric means.
Crane (1972) conducted an extensive study of 102 mathematics authors

and 221 rural sociology authors in the United States in which she used both
bibliometric methods and questionnaires to uncover emergent relationships
through sociometric methods. Based on her findings, she proposes that
scientists within a research field organize themselves into subgroups of
informal networks of personal relationships, or invisible colleges41 that are
characterized by strong ties based on informal collaboration. These
invisible colleges are then linked to individuals within other research fields
through weak ties by their members, thus facilitating the diffusion of
information both to and from each field. A common language based on a
similar orientation towards research facilitates communication between
individuals from different fields. With regard to performance, Crane
suggests that the position of a scientist in the invisible college impacts his
or her awareness of existing research as well as how rapidly he or she
obtains information. Furthermore, Crane found that productivity in terms
of innovations and publications tended to be unevenly distributed, i.e., a
small percentage of researchers were responsible for a large percentage of
innovations and publications. Finally, Crane argues that invisible colleges
have lifecycles, growing and fading depending on the state of the central
scientific research problem42.
40
A co-citation is the citation of two different publications in a third publication and
therefore is a special kind of network link between publications A and B. If the author
of publication C cites both A and B, the two must be thought to have at least something
in common. Such a co-citation implies that the authors of A and B may be studying the
same specialty and may be in communication with one another (Lievrouw et al., 1987).
41
Crane defines an invisible college as a communication network of a subgroup of
researchers within a research area (1972: 35).
42
The primary contribution of Crane’s study is that she synthesized a more coherent
understanding of the social processes that underlie the growth of science. She agrees
with Kuhn (1962) in that scientific paradigms do exhibit a lifecycle and at the same time
Following Crane, several researchers have conducted studies within the

physical and social sciences focusing on identifying invisible colleges
using either bibliometric or sociometric/anthropological methods; however,
few have combined them43. In response to this, Lievrouw, Rogers, Lowe,
& Nadel (1987) conducted an impressive study of invisible colleges among
biomedical scientists through a triangulation strategy that involved the
analysis of US-funded grants, a literature review, co-citation analysis,
questionnaires, and interviews. The 58 researchers in the study were
located primarily in the United States and communicated with one another
frequently through face-to-face meetings and telephone calls but rarely
through written means. The most interesting finding of this research is that
the communication network, or social network, among scientists was
clearly distinct from the actual content of the work in which they engaged
as based on the bibliometric methods. This finding contradicts the widely
held assumption that the social structures in science reflect in some way the
intellectual structure of the research specialty. Thus, the definition of
invisible colleges becomes critical. If we define an invisible college as a
social structure of communications, then we arrive at a different picture
than if we define an invisible college based on the content of the actual
research being performed as evidenced through bibliometric methods.
In an extensive study of the structural dimensions of scientific

communities, Schott (1988) took the center-periphery model as a starting
point and proposed structural analysis as a complement. The center-
periphery model was developed in response to the observation that
international participation in science is not equally distributed among the
participating countries. This model describes the structure of this network
as a restricted center that is the primary source of creativity, and as a result,
it influences the content and sets the direction that dominates the
intellectual work of scientists in the periphery (Schott, 1988)44. In this
study, Schott analyzed almost two million bibliographic references in the
Science Citation Index in order to investigate the underlying regional
structure and emergent social structure of scientists. In the overall center-
periphery model and consistent with previous research, Schott found the
she concurs with Price (1963) who proposes that the growth of science exhibits a
logistic curve.
43
See Lievrouw et al. (1987) for a review.
44
For further discussion on the center-periphery model, see Ben-David (1969, 1971),
Shils (1972, 1975).
80 CHAPTER THREE
United States as the pervasively influential world center. However, a

further investigation of the networks revealed that there were six regional
areas within which scientific communities strongly influenced each other.
Schott argues that the basis of influence in these regional networks is
interpersonal relationships between the scientists, promoted by collegial
and educational ties as well as geopolitical ties, such as propinquity and
language commonality.
In a recent study of a more limited nature, Tuire & Erno (2001)

investigated the formal and emergent networks of 104 professors in
education at eight Finnish universities using citation counts and sociometric
surveys measuring personal contacts. Similar to Lievrouw et al. (1987),
these researchers also found two distinct networks: individuals who were
central in the collaboration network were not necessarily those who were
central in the citation network. In addition, they found that professors were
much more likely 1) to exchange knowledge than to collaborate through
co-authorship and 2) to exchange knowledge with others in their own
university than with others outside their university. Contrary to previous
research (e.g., Crane, 1972), these researchers found a relatively thin
network of inter-university collaboration with invisible colleges forming
within universities as opposed to across them. While one explanation may
again be the importance of face-to-face, interpersonal relationships, the
authors propose that this result may also be due to the nature of education
as opposed to the more collaborative nature of the physical sciences,
indicating a relationship between structure and the underlying practice
knowledge.
On a final note and as mentioned earlier, while epistemic communities

within international relations do fall outside the scope of this thesis, they
are worth mentioning on a more general level to inform the reader.
Researchers have conducted a series of studies on epistemic communities
that describe how these communities have been influential in ascertaining
national and international policy changes and coordination, e.g., nuclear
arms control and protection of stratospheric ozone. Briefly, these studies
investigate the processes through which national and international
consensus is reached within a given domain of expertise and through which
consensual knowledge is diffused. For these studies, see the Winter Issue
of International Organization, 1992, volume 46.
3.3.2 Firm Perspective

Turning from the perspective of the scientific community to that of the
firm, Czepiel conducted one of the first studies looking at the diffusion of
technical knowledge from a sociological perspective as opposed to from a
purely rational economic perspective. Investigating the diffusion of a
major innovation (the continuous casting process) in the U.S. steel industry,
Czepiel conducted highly structured interviews of managerial and technical
people in 18 firms. These interviews revealed the existence of a
functioning inter-organizational distributed network of practice. This
network of practice linked the firms together, and respondents used this
network to gather information regarding the decision to implement the
innovation despite several barriers to such informal interaction (e.g., great
physical distances, competitive industry, avoidance of collusive activity,
etc.).
We return now to the technology transfer literature discussed above under

intra-organizational distributed networks of practice and find that there are
a number of relevant studies to inter-organizational distributed networks of
practice. Again based on Tushman’s fieldwork, Allen, Tushman, & Lee
(1979) found evidence of a relationship between project performance and
the degree to which project team members communicated with
professionals outside of their organization, e.g., researchers in universities,
contacts met at conferences, and external technical consultants. For
example, basic and general research projects showed higher performance
when all project members maintained high levels of informal technical
communication with these external contacts. However, product
development projects exhibited higher performance when external
communications were monopolized by one or a few project members, i.e.,
external gatekeepers. As an explanation, these researchers hypothesize that
research tasks are of a more universal nature than development tasks, thus
individuals working on research projects can more easily and efficiently
communicate across organizational and national boundaries than
individuals working on development projects. They base their reasoning
on Price’s (1965) proposed distinction between science and technology.
Science and scientific problems are argued to be universal, thus scientists
working within a given specialty work towards the same ends, operate
within a common social system (Kuhn, 1962; Crane, 1972) and share a
common language and set of methods. Technology, on the other hand, is
82 CHAPTER THREE
argued to be less universal since technological problems are highly

localized and defined in terms of the interests, goals, and local culture of
organization, thereby reducing the ability of individuals to communicate
across national and organizational boundaries. Thus, when individuals
communicate with others outside the organization on local problems,
outsiders have difficulty fully understanding the nature of the locally
defined problem. While both parties may think that the external individual
understands the problem, this understanding is usually incomplete and
proposed solutions or suggestions are unlikely to match the locally defined
solution space, thus resulting in poor performance.
Allen, Tushman, & Lee (1979:703) further defined gatekeepers as

“individuals who maintain consistent, ongoing contact outside their
organizations, who understand the way in which outsiders differ in
perspective from their own organizational colleagues, and who are able to
translate between the two systems”. Thus, they argue that gatekeepers are
translators, a term borrowed from Katz & Kahn (1966), and as such, they
gather knowledge external to the local group and organization, translate
this knowledge into the local language and setting, and then share it with
appropriate individuals within the local group and organization. In this
sense, gatekeepers play a similar role to Wenger’s brokers (1998) through
their participation in inter-organizational distributed networks of practice.
While the above work investigated knowledge coming into the firm,
researchers found evidence a little over a decade ago that this knowledge
flow is not unidirectional. Rather, individuals often participate in two-way
knowledge flows in which internal knowledge is traded for external
knowledge. Von Hippel (1987, 1988) performed one of the first studies
investigating this trading, in which he merely documented the phenomenon
of informal know-how trading of product and process innovations without
investigating the antecedents or outcomes of such trading. Defining know-
how trading as the “extensive exchange of proprietary know-how by
informal networks” (1987:291), von Hippel interviewed plant managers
and other managers by telephone in eleven firms in the US steel minimill
industry regarding their trading activity45. He found that in the firms
45
A precursor of von Hippel’s work was the research by Robert Allen (1983) in which
he discussed a phenomenon in the nineteenth-century English steel industry that he
called “collective invention”. Allen observed that some firms revealed competitive
information, such as new plant designs, to other firms in the industry through informal
studied, only one did not routinely trade any proprietary know-how and this
firm was considered to be an outlier in terms of know-how trading by some
of the other firms. At the ten firms that did participate, interviewees
“emphasized that they were not giving know-how away – they were
consciously trading information whose value they recognized” (1987:295,
italics in original).
Von Hippel (1987) found further anecdotal evidence of this phenomenon

when he conducted a series of pilot interviews in several US industries.
Results provide suggestive evidence that know-how trading is quite
common in some industries, such as aerospace and waferboard
manufacturing mills, while essentially absent in others, such as powdered
metals fabricators. In addition, his results suggest that this activity ranges
from being an accepted norm to being a quasi-covert activity with top
management generally not aware or approving of the activity.
In a more extensive study than von Hippel’s, Schrader (1991) surveyed by

mail 294 technically oriented middle-level managers in 127 firms from the
US specialty steel and minimill industry. Schrader finds that these
employees make their decisions to trade knowledge based on the economic
costs to the firm. He finds that the likelihood that a transfer would occur
declined 1) the more the firms were direct competitors, 2) the more difficult
it was to access the information from alternative sources, and 3) the more
highly valued the information was to the person making the transfer. While
roughly 29% of the transfers were between competing firms, Schrader
found that often information that was not related to the domain in which the
two firms competed was traded or that the information could be acquired
from another source relatively easily. As for the reasons underlying the
trades, 72% of the respondents expected that their chances of receiving
information in return would increase after the trade. Thus, Schrader argues
that it is the incremental change in the likelihood of receiving information
that is economically beneficial to the firm that is important in determining
the benefit to a transfer. Schrader then attempted to link this informal
disclosure and publication in the engineering literature. Von Hippel describes the
difference between his and Allen’s findings in the following way: “The essential
difference between know-how trading and collective invention is that know-how trading
involves an exchange of valuable information between traders which is at the same time
kept secret from non-traders. In contrast, collective invention requires that all
competitors and potential competitors be given free access to proprietary know-how”
(1987: 297).
84 CHAPTER THREE
information exchange to the economic performance of the firms. Managers

were asked to evaluate their firm’s performance relative to the industry
average on a 7-point scale (1 – well below average, 7 well above average)
in addition to the firm’s general propensity to participate in informal
technical information exchange. A positive correlation (r=0.19, p<.001)
was found between the two, providing suggestive evidence that there is a
positive relationship between informal know-how trading and firm
performance.
Kreiner & Schultz (1993) performed a small study on knowledge sharing

in the Danish R&D biotechnology industry. Based on only 16 interviews
of researchers and research directors in university and industry, they found
that individuals liberally shared knowledge that was even of a confidential
nature with others in their personal networks. In addition, they found that
successful collaboration between university and industry was often the
result of emergent personal relationships.
Perhaps inspired by the work on informal know-how trading, Macdonald

& Williams (1993) argue that the above research by Allen and colleagues
on gatekeepers did not investigate the participation of these individuals in
external knowledge exchange since these studies merely looked at the
gathering of external knowledge through oral communication channels.
Thus, these researchers conducted a limited study using a mail survey of
125 individuals working predominantly in science and engineering in the
United Kingdom. In this study, they found that individuals who were
gatekeepers within their organizations were also more likely to engage in
external knowledge trading. These individuals traded knowledge with
others with whom they had a personal relationship, and these relationships
were characterized by dyadic reciprocity.
In a further extension of their own work, von Hippel & Schrader (1996)
conducted an interesting study to investigate the possibility of managing
knowledge trading. They scoured industry to find an example of such
managed trading and found the practice of “oil scouts” in the oil
exploration industry. In this example, firms appointed oil scouts to trade
geological information on a particular well or area. Management then
mandated geologists to use the oil scouts and not to go around them by
using their personal networks even if they felt it to be more effective46.

However, von Hippel & Schrader found that geologists continued to go
around the oil scouts and participate in informal knowledge trading with
colleagues at rival firms, even though they knew it was against
management’s wishes. The argument provided was that the use of oil
scouts was ineffective. Since oil scouts were only trained finders of
information and not using the information themselves, they were unable to
understand the content of the information and thus not effective in their
actions.
In the only comparison study, Appleyard (1996) investigated the informal

knowledge sharing patterns in the steel and the semiconductor industries in
Japan and the United States. In her survey of 134 respondents of a non-
random sample47, she found only a marginal difference in the level of
overall knowledge sharing between the two industries. As for Japan vs. the
United States, respondents in both the U.S. and Japan rated colleagues in
other companies to be the most important sources of external technical
information.
Returning to the scientific world, Liebeskind, Oliver, Zucker, & Brewer

(1996) performed a small study of two US biotechnology firms, looking
only at the authors of scholarly publications as a measure of scientific
knowledge exchange. This study found that of the total of 503 publications
46
In this example, when a geologist in one firm is interested in finding some geological
information, an oil scout is supposed to be asked to find this information. In some
cases, this information is proprietary and can only be obtained from a rival firm. So, the
oil scout may approach a scout from another company and negotiate a trade. In some
cases, information is traded. However, in some cases the providing company does not
have any information that it immediately wants in return. Often the trade is still
concluded with the mutual understanding that the receiving firm “owes” proprietary
data of similar value to the providing firm. What is special about this form of trading is
that one or two trading intermediaries are placed between the individual or firm desiring
the information and the firm or individual holding the information. Traders rarely cheat
since their reputation and thus their value and jobs are dependent on their behaving
according to the rules. Thus, this trading can be said to be managed since the loss of
proprietary information without any form of beneficial return is restricted. Oil scouts
are professional traders and thus they should have a deeper understanding of the rules of
the trade. A second benefit to the system is that there is a centralization of IOUs that
helps to defer the building up of a trade imbalance between firms.
47
The author states that the respondents in the sample may be biased towards
knowledge sharing due to the selection means, i.e., participating in benchmarking study
or industry-related meeting.
86 CHAPTER THREE
during the ten-year period investigated, almost none of the 291 publications
that were co-authored by scientists working at different firms was governed
by a formal contract or other mechanism between the two firms. This
finding indicates that scientists collaborate to a high degree in inter-
organizational distributed networks of practice.
Extending Liebeskind’s research in a study of a more exploratory nature,

Oliver & Liebeskind (1998) investigated the interaction between the
organizational and individual levels of collaboration within the
biotechnology industry. Through reviewing relevant literature and
conducting an unspecified number of interviews with scientists, corporate
executives, and university technology-transfer officers in Israel and the
United States, they propose that exchanges of new scientific knowledge
occur in interpersonal relationships while formal inter-organizational
arrangements serve to support knowledge commercialization. They found
that scientists from the biotechnology firms generally did not collaborate
with scientists from other firms but with scientists at research institutions or
other universities. In addition, while there was no formal agreement,
individuals generally asked management for approval to enter external
collaborations. A final finding of interest is that scientists in large
pharmaceutical firms participated to a considerably lower level in inter-
organizational distributed networks of practice than scientists working in
new business firms. The reason provided is that few university-trained
scientists choose to work directly for these large firms for fear of being cut
off from reciprocal relationships with university researchers due to
organizational bureaucracy.
Continuing the work on knowledge exchange, Isabelle Bouty (2000)

investigated the knowledge exchange decisions of 38 R&D supervisors and
researchers in France through a case study analysis involving interviews.
In this very thoughtful study, Bouty examined the decisions regarding with
whom to exchange resources such as knowledge in informal relationships.
First, in contrast to some of the previous studies, she found that individuals
only exchanged knowledge that they did not consider to be confidential.
Second, individuals exchanged knowledge with others with whom they
were mutually acquainted, shared a high level of trust, and whom they did
not consider to be a competitor. However, Bouty makes an interesting
comment regarding confidentiality, arguing that it is socially constructed.
She writes,
“With regard to most resources, though, confidentiality is left to

the interpretation of the scientist. This interpretation results
from a personal judgment as to the interests of her or his
employing firm. It is grounded in a social context and in the
scientist’s experience in the laboratory. Thus, as personal
judgments can differ between individuals, there can be
significant variations in confidentiality appreciation.
Specifically, certain “secrets” can be common knowledge in a
community, although they officially are confidential. For
example, researchers explained: ‘If there is a promising subject,
we know that…all the companies are working on it. The
research directions are not a taboo subject between us…The
global strategy is known…there are open secrets.’” (Bouty,
2000:54).
In more European-based research, Lissoni (2001) conducted a unique study

of machine firms located in the province of Brescia, Italy that included
eight textile machinery firms, ten metalworking machine tool firms, and
seven plastic-processing machinery firms. Using data from an unspecified
number of interviews with unspecified individuals (e.g., position) at these
firms and from a questionnaire of 200 engineers engaged in design,
prototyping, and testing activities in the firms, the researcher focused a
considerable portion of the study on creating a detailed reproduction of the
design and production work-flow for each company in the sample.
However, there are some results relevant to our study. First, a surprisingly
low percentage of engineers (30%) signaled that they had any kind of
relationship (either friendship or technical) with engineers in other firms.
Secondly, only 18% of the entire sample indicated that they entered into
technical discussions with other firms’ engineers and only 4.5% indicated
that they discussed current projects. Based on these results, Lissoni argues
that knowledge does not circulate freely throughout the Brescia
geographical cluster, rather it circulates within a few smaller communities
comprising individuals linked together by ties based on trust and
reputation.
One final study worth noting here is that by Jarvenpaa & Staples (2001).
Although this study does not focus explicitly on networks of practice, it is
of particular relevance since the researchers examine attitudes towards
88 CHAPTER THREE
information and knowledge sharing and as such, it complements the above

studies on knowledge exchange. Jarvenpaa & Staples (2001) built on the
laboratory studies of Constant, Kiesler, & Sproull (1994) in which they
found that individuals had different attitudes as to whether they were
sharing information or knowledge when defined as expertise. These
laboratory studies by Constant et al. suggested that an individual’s
prosocial attitudes and norms of organizational ownership affect his or her
decision to share tangible information; however, the sharing of expertise is
influenced by personal benefits. Jarvenpaa & Staples (2001) extended
Constant et al.’s study to include contextual factors, such as information
culture and task interdependence, in an extensive study of the academic and
administrative staff of an Australian and a Canadian university. Using data
from 1125 employees (27%) from the Australian university and 810 (26%)
from the Canadian university, they investigated attitudes towards
information and knowledge sharing internally and with others in an inter-
organizational professional task force. They found that the ownership of
information and knowledge products that individuals had created was not a
zero-sum game in terms of organizational vs. individual property rights. In
other words, they argue that self-ownership coexists with organizational
ownership, and employees feel that organizations do have rights to the
labor of their employees, including their expertise. However, they did find
that individuals attached organizational property rights less to information
products than they did to their own expertise.
3.3.3 Discussion of Inter-organizational Distributed Networks of

Practice Studies
Summarizing the above studies in table 3.3, we find that while the majority
focuses on cognitive aspects, there are some relevant studies to structure
and performance. The studies on scientific communities focus on structure,
and results indicate that while individuals may cite the work of others who
may be geographically dispersed, they generally collaborate with others
with whom they interact more frequently in face-to-face settings, thus
providing further evidence of the importance of face-to-face interactions
and propinquity for the building of emergent networks. Additionally, this
research suggests that scientific communities are characterized by a core
and periphery, with the core influencing the direction of knowledge
development and thus the work of those in the periphery. With regard to
performance, these studies suggest that performance depends on an
individual’s position in the network; however, beyond this, there is little

research focused explicitly on structure.
For the studies from the firm perspective, none specifically focuses on
structure or uses social network analysis other than the studies by Allen and
colleagues. However, these studies examined primarily only the position
of the individual researcher within the team and not within the inter-
organizational network of practice as well as investigating only knowledge
flows into the firm and not out of the firm. Regarding performance, we
find a relationship between participation by a firm’s members in inter-
organizational distributed networks and performance at the firm and project
level. Evidence suggests, however, that this relationship may be dependent
on the degree to which the task knowledge is local or universal. In higher
performing projects in which knowledge is more locally defined,
gatekeepers or translators similar to Wenger’s brokers facilitate the
acquisition of external technical knowledge. However, in projects where
knowledge is more universal, higher project performance may be achieved
when all project members participate in inter-organizational distributed
networks of practice. These performance results should be regarded with
some caution since the majority of these studies were conducted thirty
years ago focusing only on researchers in R&D operations, and the more
recent study (Schrader) is limited, providing only suggestive evidence of a
positive relationship between firm performance and knowledge sharing in
inter-organizational distributed networks of practice.
An extensive number of these studies do investigate know-how trading and

knowledge sharing between members of inter-organizational distributed
networks of practice. These studies suggest that knowledge sharing within
inter-organizational distributed networks of practice is quite a common
occurrence across industries and nations with members of these networks
sharing and exchanging valuable knowledge with each other through these
emergent channels. However, the management of participation in these
networks is difficult, with individuals often making their own decisions to
participate and share knowledge without management’s consensus or even
awareness. Norms of reciprocity tend to dominate the knowledge sharing
activities by members of inter-organizational distributed networks of
practice, with individuals consciously providing knowledge to others in the
expectation that they will receive something in return. With the
questionable exception of Bouty, knowledge sharing may even include the
90 CHAPTER THREE
exchange of confidential organizational knowledge. Bouty raises a very

interesting point though - confidentiality is socially constructed, and as one
of her interviewees even noted, there are “open secrets”. Research by
Jarvenpaa & Staples further touches on this aspect of socially constructed
confidentiality since they find that the more an individual views their
knowledge as his or her personal expertise, the more the individual regards
such knowledge as his or her own property and not that of the individual’s
organization.
While these studies are of quite an international scope, they are still limited
in other areas. First, the vast majority of these studies investigate only
researchers, scientists, and engineers at universities or high technology
firms, thus providing us with little understanding of inter-organizational
distributed networks of practice in non-science based professions or at
lower levels of the firm. Second, while we do find a few studies focused
on the structural properties of these networks using social network
measures, these studies are only based on researchers and scientists and
rely heavily on publicly available data. Third and perhaps not too
surprising, all of these studies have a quantitative focus in terms of
methodology, with data collection occurring primarily through publicly
available data or surveys. Finally, we find no investigation of performance
at the individual level or the relationship between participation in different
networks of practice and performance.
Table 3.3 Selected Studies of Inter-organizational Networks of

Practice (IONP) *
Primary
Focus
Structure
Growth of Crane Bibliometrics 102 Informal
scientific 1972 and surveys mathematics collaboration
knowledge authors and between individuals
221 rural within specific
sociology research area creates
authors in US invisible colleges.
Individuals well
connected in
invisible colleges
access information
rapidly. Invisible
colleges have
lifecycles.
Structure of Lievrouw Analysis of 58 biomedical Little overlap
scientific et al. 1987 US funded scientists, between emergent
communication grants, primarily in the networks and citation
literature US networks among
review, scientists. Face-to-
bibliometrics, face meetings and
questionnaires, telephone played an
and interviews important role in
collaboration with
little use of written
means.
Structure of Schott Bibliometrics Two million US center, but six
global 1988 bibliographic regional areas in
scientific references in which local
network vs. the Science communities
regional Citation Index, influence each other,
scientific global thus indicating
networks importance of face-
to-face interpersonal,
relationships
between scientists
based on
propinquity.
92 CHAPTER THREE
Primary
Focus
Structure of Tuire & Bibliometrics 104 professors Collaboration
scientific Erno 2001 and surveys in education at networks and citation
citation vs. eight Finnish networks distinct in
collaboration universities structure of
networks individuals and
density. Invisible
colleges formed
within universities as
opposed to across
them.
Performance
Relationship R&D Case studies Primarily US Universal knowledge
between studies, with R&D more easily
external e.g., Allen, sociometric laboratories and communicated
technical Tushman, surveys their projects across national and
communication etc. (1960s, organizational
and project 1970s) boundaries than
performance more locally defined
knowledge.
Relationship
between required
task knowledge,
number of
gatekeepers, and
project performance.
E.g., higher
performing
development projects
have external
communication
monopolized by
small number of
gatekeepers.
Gatekeepers similar
to Wenger’s brokers.
These individuals
generally in
management
positions or highly
technically
competent and seen
as experts.
Relationship Schrader Mail survey 294 technically Suggestive evidence

between know- 1991 oriented of positive link
how trading middle-level between know-how
and firm managers in US exchange and firm
performance specialty steel performance.
and minimill Individuals base
industry decisions to trade
know-how on
economic costs to
firm. Norm of
reciprocity in dyadic
relations.
Cognitive
Aspects
Diffusion of Czepiel Interviews, Managers and Decision-makers use
technical 1975 Social technical IONP to gather
innovation network people in 18 knowledge regarding
analysis US steel firms implementation of
innovation
External von Hippel Telephone Unspecified Informal know-how
knowledge 1987 interviews number of trading is norm
exchange and managers in 11 within minimill
knowledge US steel industry. Anecdotal
sharing minimills evidence that trading
and covertness of
trading varies across
industries.
Informal Kreiner & Interviews 16 researchers Liberal trading of
collaboration Schultz in Danish R&D confidential know-
between 1993 biotechnology how. Emergent
scientific firms and relationships often
organizations universities led to successful
formal collaboration
between university
and firm.
Participation Macdonald Mail survey 125 individuals Gatekeepers
of gatekeepers & Williams in science and participate in dyadic
in external 1993 engineering in reciprocal
knowledge UK knowledge exchange
exchange in external networks
of practice.
94 CHAPTER THREE
Can external von Hippel Case study Oil scouts in oil Difficult to manage
knowledge & Schrader industry know-how trading
exchange be 1996 through formal
managed? means. Sharing of
tacit knowledge
dependent on
common
understanding of
work.
Comparison of Appleyard Surveys 134 engineers, Little differences in
patterns of 1996 sales, and informal external
external quality control knowledge sharing
knowledge individuals in between industries.
sharing US and External colleagues
Japanese steel most important
and source of external
semiconductor technical
industry information.
Means with Liebeskind Bibliometrics 503 External
which new et al. 1996 publications in collaboration in
biotechnology two publications by
firms source biotechnology individuals in firms
scientific firms in US not governed by
knowledge formal contract or
agreement.
Interaction Oliver & Interviews Scientists, Scientists in biotech
between Liebeskind corporate firms generally
individual and 1998 executives, collaborate only with
organizational university scientists in research
levels in technology- institutes or
external transfer officers universities and not
scientific in Israel and with individuals in
collaboration US biotech other firms.
industry Collaborations not
governed by formal
agreement, but
subject to informal
management
approval.
Individual Bouty 2000 Case study 38 R&D Three-step decision
decisions to with supervisors and model to exchange
externally interviews researchers in knowledge in inter-
exchange France organizational
strategic distributed network
knowledge of practice.
Confidential know-
how not exchanged.
External Lissoni Interviews 200 electrical Low percentage of

knowledge 2001 and surveys and mechanical engineers discussed
sharing in engineers in 8 current projects
geographically textile (4.5%) or other
co-located machinery technical issues
cluster firms and their (18%) with other
customers, 10 firms’ engineers in
die-casting the province.
mold Knowledge did not
producers, and circulate freely in the
7 plastic- cluster, rather it
processing circulated within a
machinery few inter-
firms in Italian organizational
province distributed networks
of practice.
Knowledge Jarvenpaa Surveys 1125 Self-ownership and
ownership & Staples employees organizational
2001 (27%) from ownership of
Australian and knowledge coexist.
810 (26%) from Personal expertise
Canadian felt to be more of
university individual property
than tangible
information product
owned by
organization.
3.4 Studies of Electronic Networks of Practice
Before starting this review of the last type of network of practice, it is

necessary to note how we have limited this review. Due to the rapid
growth of internet usage, we are beginning to see a significant number of
impressive studies on online communities. For example, in one of the first
large-scale web surveys, Wellman, Quan-Haase, Witte, & Hampton (2001)
conducted an extensive study of the impact of the internet and participation
in electronic communities on society’s social capital through a survey of
39,211 visitors to the National Geographic Society website. While at first
glance many of these studies appear relevant to our task at hand, further
investigation reveals that the majority of these do not revolve around a
specific work practice. Thus, in order to keep the review focused on
networks of practice, we do not review studies of online communities in
96 CHAPTER THREE
education, the body of research on Usenet newsgroups more oriented

towards social support and self-help (with one exception), or studies
looking at the effect of electronic community participation on society. Nor
have we included studies looking at the human computer interface aspects
of online conversation, such as that of chat rooms48, or on internet or media
usage in general. Making these delimitations leaves us then with only a
handful of studies that are of interest due to their explanatory power
regarding electronic networks of practice from a firm’s perspective.
However, as you will see, some even stretch our definition of an electronic
network of practice somewhat. A final point is that due to this small
number of studies, we include both intra-organizational and inter-
organizational electronic network of practice studies here.
The first study we report is an inspiring study by Constant, Sproull, &

Kiesler (1996) that has received considerable recognition due to its early
appearance and impressive scope. These researchers conducted a study of
a broadcast mailing list (i.e., listserv) for the entire employee body of more
than 11,000 employees at Tandem Computers, Inc. This mailing list was
used for work-related broadcast messages for the entire organization,
including announcements from headquarters, industry news, and requests
for information. The particular focus of this study was on broadcast
requests for information of any kind as long as it was work-related, which
were about 30% of all the postings on this list during the six-week research
period. As such, this mailing list is on the border of being considered an
electronic network of practice since it contained all 11,000 employees and
questions could revolve around any form of work practice. However, this
study does provide some insights into an intra-organizational electronic
network. Using data from surveys sent to message posters (55 information
48
Research has found that while different kinds of electronic communities share some
characteristics, they may also be distinctive. For example, support groups are
distinctive in their use of electronic group membership as a legitimating strategy and in
their use of both expertise and personal experience as warrants for advice (Galegher,
Sproull, & Kiesler, 1998). For a review of research on electronic communities within
the educational field, see Johnson (2001). For a study of 155 cultural newsgroups (e.g.,
African-American, Pakistan, etc.), see Choi & Danowski (2002). For a study on
conversation in chat groups, see Donath, Karahalios, & Viégas (1999). See Lueg &
Fisher (2003) for numerous studies looking at online social spaces. For how to manage
electronic communities from a practitioner’s viewpoint, see Williams & Cothrel (2000).
A number of books also focus on building and facilitating electronic communities, for
example Kim (2000) and Palloff & Pratt (1999).
seekers and 295 information providers) and text analysis of the postings to
the mailing list during the six weeks, the researchers found that people
helped others primarily out of organizational commitment and norms of
reciprocity and because they enjoyed solving problems and helping others.
The researchers determined this mailing list to be effective since message
posters received useful technical advice - 49% of the respondents indicated
that the replies had solved their problem. As for the relationships between
individuals, 81% of the message respondents said that they did not know
the message posters at all. In addition, they find that similarity in terms of
managerial status, hierarchical level, firm experience, or industry
experience does not explain the interactions and that the pattern of
interactions is characterized by generalized reciprocity and not dyadic
reciprocity. Finally, the usefulness of the replies has no correlation with
the number of replies given, thus providing support for Granovetter’s
(1983) and Burt’s (1992) notions that weak ties are useful if they bridge
areas of superior resources.
Extending the experiments on information sharing on students by Constant

et al. (1994) mentioned above in this chapter to include contextual factors
such as information culture and task interdependence, Jarvenpaa &
Staples (2000) conducted a large study of the determinants of the use of
electronic collaborative media (e.g., listservs, email, web-browsers) by
individuals in the workplace. Questionnaire responses were collected from
1125 academic and administrative staff (27% response rate of all academic
and administrative staff) in an Australian University. Although the
questions relating to media usage did not specifically ask whether these
collaborative media were used for conducting work-related tasks, they do
provide an indication of collaborative media usage for knowledge gathering
and sharing with other individuals in intra-organizational and inter-
organizational networks of practice. Contrary to expectations, this study
found a negative relationship between an open, organic information culture
and the use of the collaborative media. Possible explanations put forth by
the authors are that in this organization, hierarchy may require more use of
the electronic media or that individuals use these media to circumvent the
closed information culture to access knowledge that is not freely shared.
Another finding of interest is the negative relationship between the use of
collaborative media and organizational information ownership, which
supports Constant et al.’s (1994) research.
98 CHAPTER THREE
In a well-known study similar to Constant et al. (1996) that is available as

an MIT working paper online, Lakhani & von Hippel (2000) examined
the Apache Usenet discussion group through an analysis of website log
data over a four year period and a survey of 366 participants. They found
that information providers were relatively concentrated with 50% of the
answers provided by the 100 most prolific providers, representing 2% of all
providers. In contrast, 24% of the information seekers asked 50% of the
questions. They also found that individuals were motivated to share by
reciprocity, interest in advancing the community, intrinsic rewards, and
increased reputation. Providers strongly disagreed with the statement “it is
part of my job”, indicating that helping was indeed discretionary.
In one of the first published studies on inter-organizational electronic

networks of practice, Wasko & Faraj (2000) conducted a somewhat
exploratory study similar to that of Lakhani & von Hippel. They
investigated participation in three technical usenet newsgroups (computer
language C++, computer objects, and computer database) through an email
survey of 342 participants and content analysis of the survey’s open-ended
responses. Looking first at participation in general in the community, they
found that these communities appeared to have certain norms, such as that
individuals only help others who first try to help themselves. Prestige also
seems to play a role, with individuals supporting and even attacking each
other. Finally, they found suggestive evidence of a negative relationship
between network size and its ability to create value for its members.
Turning to motivation for participation, Wasko & Faraj found that
individuals were motivated by both self-interest and collective interest,
where self-interest included both tangible returns (e.g., efficiency in
completing work tasks) and intangible returns (e.g., intrinsic satisfaction)
and collective interest included care for the community (von Krogh, 1998),
prosocial behaviors (e.g., “the right thing to do”), and an interest in
advancing the community. However, the most frequent response was
reciprocity - that individuals felt that they should give back to the
community – thus reflecting generalized as opposed to dyadic reciprocity.
These findings further support the view of knowledge as a public good in
electronic networks of practice.
Molly Wasko (2001) continued the work from the previous study in her
impressive doctoral dissertation on electronic networks of practice.
Through interviews, content analysis of messages posted during a two-
month period, and survey data collected from 160 respondents in an inter-
organizational electronic network of practice of a US professional legal
association, she examined the underlying personal and social factors that
predict knowledge contribution in electronic networks of practice. Her
results are partially consistent with the previous research mentioned above
since she found that people exchanged knowledge with strangers in the
electronic network of practice based on the expectation of generating some
type of return. Respondents contributing knowledge to the network did not
expect to receive tangible returns, rather they were interested in enhancing
their reputation in the network as well as reaping other intrinsic returns
such as the desire to challenge themselves. In addition, the results revealed
that active responders did not behave altruistically, and the only significant
difference between responders and non-responders was that responders
desired to receive intrinsic returns in the form of enjoyment and challenge.
Based on these findings, Wasko argues that knowledge is best
characterized as a private good in this context, where people engage in its
exchange in order to receive commensurable benefits. However, people
are still willing to engage in the provision of public information goods.
The implication is that the expectation of returns does not necessarily
translate into whether the public good will be provided, but it does seem to
have implications for the quality or “helpfulness” of the good. We further
develop her thinking in Article 2 of this thesis.
Additionally, through content analysis, Wasko found that this particular

electronic network of practice exhibited the norm that individuals only help
others who first try to help themselves. Individuals who had “done their
homework” and then asked questions almost always received helpful
answers. Regarding message content, the majority of postings were
knowledge related, had a very low content of socializing, and were
respectful and encouraging, with only a very few containing flames or
scathing remarks. Finally, knowledgeable members generally corrected
postings when appropriate. However, interviews revealed that respondents
were of the perception that mentor experts had migrated to their private
personal networks for knowledge exchange due to the network being taken
over by “newcomers”. One highly skilled and knowledgeable interviewee
had stopped participating in the electronic network since “it opens the
floodgate for people who are seeking advice” while others had lost interest
due to the novice level of the questions (p. 80).
100 CHAPTER THREE
Two very unique studies by Nonnecke & Preece (2000, 2003) are also
worth reporting since they are the only ones to our knowledge that
investigate lurkers in listservs. In a large demographic study (2000), these
researchers analyzed 147,946 messages posted during a twelve-week
period on 109 health and software support distribution lists. They found
that for all the distribution lists, 56% of the listserv members made no
postings while 81% made less than two postings per month during the
twelve weeks, indicating that “lurkers” made up the vast majority of
electronic community participants. However, they also found significant
differences in lurker demographics between the two kinds of distribution
lists. In a very limited study investigating why lurkers lurk (2003), ten
interviews of electronic network participants revealed that lurking was a
complex process to understand. A notable finding is that although the
interviewees did not publicly participate in the online discussions, they felt
a commitment to the network of practice and some even side-posted to
provide support to other members. Thus, these researchers propose that
lurkers should not be labeled as free-riders since if they were, then they
pose the question, “How do online groups survive in the face of almost
universal free-riding”. Thus, they suggest that lurkers be labeled non-
public participants.
Returning to Hara’s (2000) work discussed above, her investigation of the

electronic network of practice of lawyers in Indiana produced findings
similar to Wasko’s study (2001). Hara found that the more experienced
attorneys relied to a higher degree on people and face-to-face interactions
than on electronic communications or the electronic network of practice.
The electronic network consisted to a high degree of younger, less
experienced attorneys, and ties within this network were found to be
weaker than within the communities of practices that developed in co-
located settings. These findings suggest that there is a relationship between
demographics and the participant composition of different networks of
practice. One interesting finding is that participation in an electronic
network of practice fostered interaction between attorneys in face-to-face
settings. Online discussions extended into face-to-face discussions,
especially since postings were not anonymous. Finally, Hara’s findings
support previous research that electronic networks of practice support the
sharing of tacit knowledge to a lower degree than communities of practice.
3.4.1 Discussion of Electronic Networks of Practice Studies

In conclusion, while we find that there is a growing body of research on
online communities, there is scant research conducted specifically on
electronic networks of practice, be they intra-organizational or inter-
organizational. For example, no attention is paid to intra-organizational
electronic networks of practice other than Constant et al.’s related study
(1996). Interestingly, the two more extensive studies on inter-
organizational electronic networks of practice both consist of US lawyers,
and the two more exploratory studies consist of only software
programmers. Moreover, as seen in table 3.4, there are no studies
specifically relating to structure or performance. However, several studies
do suggest that interactions between electronic network of practice
members are of a generalized and not a dyadic nature. In these studies,
researchers have generally investigated individual motivations behind
participation and knowledge sharing, revealing that individuals share
knowledge with “strangers” due to expectations of returns for themselves
(e.g., increased reputation, enjoyment, etc.) as well as for the network
(advancing the community). Research by Lakhani & von Hippel further
indicates that individuals make discretionary choices regarding their
willingness to share knowledge and help others in the network.
Additionally, these studies suggest that the lens of collective action with
the knowledge of the electronic community as a public good is a useful
means of understanding these networks.
Since these inter-organizational electronic networks of practice are

conducted in internet space, we know little about the national
demographics of these individuals other than that they understand and can
write English. However, due to the global reach of the internet, it can be
surmised that participants are globally dispersed. Finally, these studies
have primarily relied on surveys and content analysis of posted messages.
102 CHAPTER THREE
Table 3.4 Selected Studies of Electronic Networks of Practice (ENP) *

Primary
Focus
Structure
------------
Performance
------------
Cognitive
Aspects
Predictors of Constant Email Mailing list Individuals generally
usefulness of et al. 1996 survey including all 11,000 strangers. No correlation
technical employees of between demographics
advice Tandem and participation.
provided in Computers. Interactions characterized
mailing list Surveys of 55 by generalized reciprocity.
information seekers Resources of network
and 291 participants more
information important than network
providers size in usefulness of
network. Individuals
provide advice due to
organizational
commitment, reciprocity,
and enjoyment.
Individual Jarvenpaa Mail survey 1125 academic and Negative relationship
factors & Staples administrative staff between electronic media
underlying 2000 in Australian use and open, organic
usage of University information culture and
electronic organizational information
media ownership.
Motivation for Lakhani & Content Four year website Sharing based on
helping others von analysis and log data and 366 reciprocity, interest in
and sharing Hippel email survey participants in the community advancement,
knowledge in 2000 Apache Usenet intrinsic rewards, and
open source discussion group increased reputation.
electronic Choice to help others was
community discretionary. Information
providers are relatively
concentrated but
information seekers are
relatively dispersed.
Motivation for Wasko & Content 342 respondents in Motives include prosocial
helping and Faraj 2000 analysis of three technical behavior, care in
sharing open-ended usenet newsgroups: community and
knowledge responses on computer language generalized reciprocity.
with others in email survey C++, computer Support for knowledge as
inter- objects, and a public good.
organizational computer database
electronic
network of
practice
Predictors of Wasko Mail survey 160 respondents Share knowledge
knowledge 2001 and content and 2,496 messages primarily due to intangible
contribution analysis of an inter- returns (e.g., reputation,
in electronic organizational US enjoyment).
network of professional legal Knowledge viewed as both
practice association private and public good.
Demographics Nonnecke Content 147, 946 messages Vast majority of members
of lurkers in & Preece analysis on 109 health and (81%) posted less than 2
online 2000 software support messages during 12-week
communities listservs period.
Motivation for Nonnecke Interviews 10 lurkers in US Lurkers provide support to
lurking & Preece other network members
2003 through side-posting.
Role of Hara 2000 Ethnography ENP of lawyers in Participant demography
information state of Indiana differs between types of
technology in network of practice. CPs
knowledge support sharing of tacit
sharing and knowledge to higher
construction degree than ENP. ENP
instigates interaction
between co-located
individuals.
104 CHAPTER THREE
3.5 Summary of Previous Empirical Studies of Networks of

Practice
In summary, we have found and reviewed some fifty studies within or

closely related to the network of practice field. We have broadly
summarized these studies in table 3.5 according to their findings relating to
structure, performance, and cognitive aspects as well as according to
methods and samples. We discuss this summary below.
Looking first at methods and samples, we find that researchers have

primarily conducted their studies of communities of practice through
extensive ethnographies and case studies that often formed the basis of
doctoral dissertations and that focused on lower level employees and non-
science based professionals located in the United States. Turning to intra-
organizational distributed networks of practice, we see that there are very
few in-depth quality studies that are specifically focused on these networks.
Here we also find a distinct and interesting change in methodology and
samples, with researchers primarily using quantitative methods involving
questionnaires and sociometric analyses while focusing on professionals
such as researchers and lawyers, again within the United States. This
choice of methodology and samples is somewhat repeated when we look at
the considerable amount of research conducted on inter-organizational
networks of practice, with researchers using questionnaires and sociometric
analyses to investigate the networks of researchers, scientists, and
engineers at universities or high-technology firms. However, contrary to
the previous studies, studies of inter-organizational networks of practice are
of a more international scope. Turning to the final category, electronic
networks of practice, we find that there is a dearth of studies focused on
this type of network of practice. Only a handful of these studies are truly
focused on electronic networks of practice with the more extensive ones
limited in terms of samples, focusing only on software programmers and
lawyers who speak English, and methodology, involving primarily surveys
and content analysis. As a result, we may clearly question the
generalizability of the research within each type of network of practice to
other settings. For example, while there is extensive research on
communities of practice on non-professionals in the United States, research
has failed to show whether we may generalize these results to professionals
or to national cultures outside the United States.
Table 3.5 Summary of Selected Empirical Studies

Notes (Primarily Relating
Network Cognitive Methods and to Structure and
of Practice Struct. Perf. Aspects Samples Performance Findings)
Communities ** ** **** - Ethnographies, -Studies discuss only
of practice case studies impact of formal
- Lower level organization on
employees, community of practice
craftsmen, non- formation and very general
professionals structural characteristics
- US oriented -No rigorous studies
specifically focused on
performance
Intra- *** *** ** -Questionnaires -Studies primarily focused
organizational with on project performance.
distributed sociometric More relevant studies on
networks of analyses performance and structure
practice -R&D mostly performed twenty
researchers and to thirty years ago and not
professionals explicitly on intra-
-US oriented organizational distributed
Inter- * ** **** -Bibliometrics, - Studies focused primarily
organizational case studies, on project performance.
distributed interviews, Performance studies of
networks of surveys, social significance conducted
practice network analysis only on scientists and
-Scientists, researchers and performed
researchers, thirty years ago.
engineers -Structure studies
-International performed only on
sites scientists and researchers
Electronic --- * *** -Content -Studies primarily on
networks of analysis and inter-organizational
practice surveys electronic networks of
-Software practice
programmers -No studies focusing
and lawyers specifically on structure or
-English performance
speaking
--- - No relevant findings or studies
* - Limited research of more exploratory or less rigorous nature
*** - Extensive research on one type of occupation (e.g., scientists & researchers)
***** - Extensive research on numerous types of occupations
106 CHAPTER THREE
Making broad comparisons of these networks in terms of the research

themes in these empirical studies provides us with the ability to easily
identify the research gaps and assumptions made within each type of
network. First, only the studies on communities of practice have focused
considerably more in terms of depth and scope on the cognitive aspects
than the other networks of practice in general. Through these studies, we
have a deep understanding in terms of identity and knowledge sharing and
the importance of physical co-location for communities of practice;
however, the lack of ethnographic research on non-co-located networks of
practice, be they inter or intra-organizational, limits our understanding of
these cognitive elements.
A second area worth noting is that if we take a step back and take a broad
look at the field, we find in our review of the empirical studies that
researchers generally depart from the assumption either that the individual
network member has already made the decision to participate in the
network or that individuals are willing to freely share their knowledge with
other network of practice members. For example, the community of
practice newcomer mutually engages with other members in order to
become a full member of the community of practice or the electronic
network of practice member has already signed up on the mailing list.
However, a few studies on electronic and inter-organizational distributed
networks of practice provide evidence that individuals do make
discretionary choices regarding the degree to which they participate and
share knowledge in a particular network of practice and that they base these
decisions on the expectation that they will receive some tangible or
intangible benefits in return.
Furthermore, this research provides some interesting perspectives with

regard to the knowledge that is shared within these networks. For example,
studies on inter-organizational distributed networks of practice suggest that
the confidentiality and ownership of knowledge is socially constructed by
networks of practice members while studies on electronic networks of
practice have opened our eyes to the examination of knowledge and the
network using the lens of collective action and public goods. Finally, in
general, we find that there really is a dearth of studies on networks of
practice from a business firm’s point of view with some areas wide open
for study, e.g., intra-organizational networks of practice, be they distributed
or electronic.
With regard to structure, as can be seen in table 3.6, we find that no

research has specifically focused on the structural properties of
communities of practice or electronic networks of practice. Looking at
research on intra-organizational and inter-organizational distributed
networks of practice, we find that the majority of these studies investigate
only the structural properties of science-based networks. Additionally, the
research by Allen and colleagues was conducted thirty years ago in the pre-
internet era. Perhaps this lack of research focusing on structure is not too
surprising given the difficulty in tracking these invisible networks.
Traditional means of collecting social network data are extremely time
consuming and analysis of the data can be quite intensive, not to mention
that access to organizations is difficult due to certain ethical and privacy
issues. However, based on these findings, we may draw some general
conclusions regarding structure in networks of practice. Results suggest
that the formal organization may influence the structure of a network of
practice, that there is a negative relationship between physical distance
between individuals and the creation of emergent relationships due to the
importance of face-to-face interactions, that structures are characterized by
a core and periphery, and that reciprocity moves from being a dyadic
exchange in communities of practice to of a more generalized form in
electronic networks of practice.
Table 3.6 Summary of Findings from Selected Empirical Studies

Relating to Structure
Primary
Research Focus Study Methods and Sample Primary Findings
Communities of
Practice
Relationship Lave & Ethnography of small Structure is more an
between learning Wenger groups of Mayan midwives, adaptive outcome of
and social situations 1991 Liberian tailors, US non- action rather than a
drinking alcoholics, precondition within
butchers in US a social system.
supermarkets, and US navy
quartermasters
108 CHAPTER THREE
Primary
Relationship Wenger Ethnography of around Structure of
between learning 1998 twenty claims processors community of
and social situations co-located in one US practice consists of
company core of full
participants
surrounded by
layers of peripheral
members.
Management of Schenkel Case study involving Formal organization
unexpected 2002 interviews, sociometric can influence CP
deviations in work questionnaire, and archival structure.
procedures data review of major
Scandinavian construction
project involving 137
managers and support
individuals
Intra-
organizational
Distributed
Networks of
Practice
Structure of weak Friedkin Sociometric questionnaire The stronger the tie
ties 1980 of scientists in seven between two
biological departments in individuals, the
US university higher the overlap
of their contact
circles.
Factors affecting Han 1996 Sociometric questionnaire Formal
creation of of 76 employees in large US organizational
emergent retail corporation boundaries affect
relationships building of intra-
organizational
relationships
Collective action Lazega Case study involving Relationships based
among individuals 2001 sociometric data of 71 on advice, goodwill
equal in power lawyers in three offices of (co-workership),
US law partnership and friendship
create social niches.
Inter-
organizational
Distributed
Networks of
Practice
Growth of scientific Crane 1972 Bibliometrics and surveys Informal
knowledge of 102 mathematics authors collaboration
and 221 rural sociology between individuals
authors in the US within specific
research area creates
invisible colleges.
Structure of Lievrouw Analysis of US funded Little overlap
scientific et al. 1987 grants, literature review, between
communication bibliometrics, collaboration
questionnaires, and networks and
interviews of 58 biomedical citation networks
scientists, primarily in the among scientists.
US Importance of
propinquity.
Structure of global Schott Bibliometrics using two Importance of face-
scientific network 1988 million bibliographic to-face
vs. regional references in the Science relationships.
scientific networks Citation Index, global Core/periphery
structure.
Structure of citation Tuire & Bibliometrics and surveys Collaboration
vs. scientific Erno 2001 of 104 professors in networks and
collaboration education at eight Finnish citation networks
networks universities distinct in structure.
Importance of face-
to-face
relationships.
Electronic
Networks of
Practice
What predicts Constant et Email survey participants of Interactions
usefulness of al. 1996 mailing list including all characterized by
technical advice 11,000 employees of generalized
provided in mailing Tandem Computers. reciprocity.
list? Surveys of 55 information
seekers and 291 information
providers
110 CHAPTER THREE
Motivations for Lakhani & Content analysis and email Information

helping others and von Hippel survey of four-year website providers are
sharing knowledge 2000 log data and 366 relatively
in open source participants in the Apache concentrated but
electronic Usenet discussion group information seekers
community are relatively
dispersed.
Finally, while significant interest in networks of practice is rooted in the

belief that there is a positive relationship with performance, as can be seen
in table 3.7, we really have very limited support for this relationship, be it
on the individual, group, or organizational level. For example, we find that
despite considerable claims by community of practice scholars that there is
a relationship between communities of practice and performance, we find
no rigorous studies other than Snyder’s somewhat related study that truly
support this relationship at any level. Additionally, most of the studies on
intra-organizational and inter-organizational networks of practice that are
of a more serious academic nature were conducted thirty years ago and
only within a few limited sites. Finally, there are no studies providing any
kind of evidence of a relationship with performance within electronic
networks of practice nor are there any studies specifically focused on
individual performance within any type of network of practice. However,
the findings do suggest that an individual’s position in a network of
practice has an impact on performance.
Table 3.7 Summary of Findings from Selected Empirical Studies

Relating to Performance
Primary
Communities of
Practice
Relationship Snyder Case study: observations, Communities of
between 1996 interviews, archival data practice influence
organizational review of US office workers organizational
learning and in two divisions of four performance
organizational offices in US Veterans through impact on
performance Benefits Association (100 to organizational
250 employees in each knowledge and
office) learning.
Relationship Lave & Ethnography of small High performers of

between learning Wenger groups of Mayan practice are full
and social situations 1991 midwives, Liberian tailors, participants of
US non-drinking alcoholics, community
butchers in US
supermarkets, and US navy
quartermasters
Relationship Wenger Ethnography of around Core members are
between learning 1998 twenty claims processors high performers of
and social situations co-located in one US CP tasks.
company
Relationship Schenkel & Survey of Major Limited support for
between community Teigland Scandinavian positive relationship
of practice and 2002 construction project between CPs and
organizational involving 137 managers and learning curve
performance support individuals. improvement.
Intra-
organizational
Distributed
Networks of
Practice
Relationship R&D Case studies with Project performance
between information studies, sociometric dependent upon
flows and project e.g., Allen, surveys of primarily US project task
performance Tushman, R&D laboratories and their knowledge and
etc. (1960s, projects number of
1970s) gatekeepers.
Knowledge Hansen Case study with sociometric The more central
integration across 1996, 1999 surveys of 120 R&D the R&D team in
subunits in multiunit projects within one US the organizational
firm multinational network in terms of
team’s unit
possessing relevant
expertise, the easier
the network search,
and the faster the
completion time.
Collective action Lazega Case study of 71 lawyers in Social pressure is
among individuals 2001 three offices of US law applied by others
equal in power partnership within social niches
to maintain
individual
performance levels.
Individuals who are
most sought out for
collaboration and
112 CHAPTER THREE
advice by others
across the firm earn
more money for the
firm.
Relationship Lesser & Interviews of five to ten Support for positive
between networks of Storck members of seven IANPs relationship
practice and 2001 comprised of knowledge between
creation of workers in a variety of large participation in
organizational value corporations (majority IANPs and
likely in US) organizational
performance.
Inter-
organizational
Distributed
Networks of
Practice
Relationship R&D Case studies with Relationship
between external studies, sociometric between required
technical e.g., Allen, surveys of primarily US task knowledge,
communication and Tushman, R&D laboratories and their number of
project performance etc. (1960s, projects gatekeepers, and
1970s) project
performance.
Relationship Schrader Mail survey of 294 Suggestive evidence
between know-how 1991 technically oriented middle- of positive link
trading and level managers in US between external
organizational specialty steel and minimill knowledge
performance industry exchange and firm
performance.
Electronic
Networks of
Practice
--------------
Having reviewed the empirical studies to date, we now move to the next
chapter in which we build on the research gaps and assumptions revealed in
this review to develop the two research purposes for the empirical studies
of this thesis.
CHAPTER FOUR
Development of Research Purposes
IN THIS CHAPTER, we develop the two central research purposes of this

thesis. These two research purposes complement previous research since
they emerge directly from the gaps revealed in the literature review in the
previous chapter: (1) structural dimensions of networks of practice and (2)
performance in networks of practice. Additionally, these research purposes
serve to fill the gaps in empirical research on intra-organizational
distributed networks of practice as well as intra-organizational and inter-
organizational electronic networks of practice.
4.1 Research Purpose 1: The Structural Dimensions of

If we return to our initial discussion of networks of practice in Chapter

Two, you will recall that social relationships between individuals are the
basis for all networks of practice. For example, it is through interactions
with others that novices become full participants and construct their
identity in communities of practice and that individuals share knowledge
through electronic communities. This is mirrored in the empirical studies
presented in the review in Chapter Two in which researchers have focused
on investigating social interactions and the related cognitive aspects.
However, studying these cognitive aspects provides only a partial
understanding of these networks.
In contrast to other areas of the social sciences that have tended to study the
“attributes” or the characteristics of individuals, groups, and organizations,
researchers within social network analysis have been paying increasing
interest to the relations between individuals, groups, and organizations
within the past few decades. In social network theory, researchers have
114 CHAPTER FOUR
found that the interactions between individuals within emergent groups

create patterns of relationships that in turn constitute the structure of the
network (Brass, 1985; Krackhardt & Porter, 1985; Burkhardt & Brass,
1990; Krackhardt, 1991). These relations may then be characterized by a
number of important properties such as frequency, stability, transitivity,
reciprocity, and multiplexity, and they facilitate the structural analysis of
social groups (Monge & Contractor 2003)49. As stated in Chapter One,
structure within social network theory is defined as the presence of regular
patterns or regularities in relationships (Wasserman & Faust, 1994) that are
represented by networks comprising sets of nodes and sets of ties depicting
the interconnections between the nodes (Wellman & Berkowitz, 1988).
This definition of structure is distinguished from others in the social
sciences since it focuses on the “concrete social relations among specific
social actors” (Wellman & Berkowitz, 1988:5, italics in original) as
opposed to other aspects such as symbols, meanings, norms, values, and
role expectations (ibid, Scott, 1998). In addition and as mentioned above,
this network definition of structure is in strong contrast to the definition of
formal organizational structure that generally refers to the prescribed
framework focusing on the differentiation of positions, the formulation of
rules and procedures, and prescriptions of authority within an organization
(Ranson et al., 1980)50.
Within social network theory, individuals and their actions are viewed as
interdependent rather than independent, autonomous units since individuals
are embedded in networks of relationships (Berkowitz, 1988; Wasserman
& Faust, 1994). As indicated above, the fundamental principle then is that
pair-wise relationships among individuals link to form networks whose
structural characteristics both are the result of dynamic interaction
processes and have an effect on individual and group outcomes. At the
individual level, due to embeddedness, individuals are involved in
multilateral resource interdependencies in which relationships provide
access to key resources, such as goodwill, advice and social support, as
well as the ability of the individual to influence or be influenced by others
(Lazega, 2001). Thus, a person's position in a network may result in both
49
For a review of social network concepts and principles, see Monge & Contractor
(20003) and Wasserman & Faust (1994).
50
As mentioned previously, interest in this formal view of structure has been heavily
influenced by Weber (1946), e.g., Hall (1963), Pugh et al. (1968,1969), and Child
(1972).
DEVELOPMENT OF RESEARCH PURPOSES 115
constraints and opportunities for the individual (Burt, 1992). At the group
or network level, there is the holistic notion of emergent properties that
suggests that at least some properties and outcomes of a social network are
a function of its complete structure and are not reducible to either an
individual actor or a single link (Degenne & Forsé, 1993). Furthermore,
some researchers argue that emergent network structures better explain
organizational behavior than formal structures (Bacharach & Lawler, 1980;
Krackhardt & Hanson, 1993; Monge & Contractor, 2003).
Social network analysis is then an analytic technique that researchers use to

represent the relational data of networks and to investigate the nature and
properties of these relations. With the exception of the studies mentioned
above, such as Crane’s work on scientific communities, the study of
lawyers by Lazega, and the studies in R&D such as those by Allen and
Hansen, to the best of our knowledge there are few other studies that focus
on investigating the structure of networks of practice from a social network
perspective. This seems surprising for several reasons. First, as discussed
further below, the logic of networks of practice carries with it strong
parallels to the structural characteristics of embedded networks. Second,
many of the arguments of the network of practice literature are based upon
the underlying principle of social network theory: the assumption that
individuals are embedded in networks of social interactions that shape their
behaviors. Looking at the extensive stream of social network literature, we
find that there is a wide range of analytical tools that describe and analyze
emergent structures. Thus, an application of social network theory and
social network measures to networks of practice would improve both our
understanding of and our ability to theorize regarding these emergent
networks. Against the background of the above discussion, the first
overarching purpose of this thesis is the following:
Research Purpose 1: To describe the structural properties of

networks of practice through the application of social network
analysis.
This purpose is further broken down into two sub-purposes. Since the
studies using social network analysis mentioned above focus on intra-
organizational and inter-organizational distributed networks of practice, we
have chosen to examine the structures of two networks of practice that have
not received any attention to date: a community of practice and an
116 CHAPTER FOUR
electronic network of practice. We discuss each of these sub-purposes in

turn.
4.1.1 Research Purpose 1a: Structural Dimensions of Communities of

Practice
We turn first to the area of communities of practice. With regard to the
connection between communities of practice and social networks, we argue
in Chapter Two that every community of practice consists of a network, but
not every network forms a community of practice. If we agree with this,
then the question arises as to whether there are specific structural properties
that are likely to distinguish a community of practice from other networks.
As mentioned above, researchers have begun to look at the various
structural aspects of communities of practice, and in particular that of
participation levels within communities. However, as far as we have been
able to discern, the participation levels as described by the initial
proponents of communities of practice, e.g., Lave and Wenger, are the only
structural aspects of communities of practice that have been discussed in
any significant detail.
Thus, in order to examine the structural properties of communities of

practice, we turn to the wide range of extensively used concepts, measures,
and techniques offered by social network analysis to find those that are best
able to describe communities of practice. We may then synthesize these
social network analysis concepts with existing concepts from the
community of practice literature to conceptually develop structural
properties distinct to communities of practice. These structural properties
may then help to detect and analyze communities of practice within
organizations, to track their development over time, or to measure their
relationship with organizational performance.
In order to fill a second gap in the literature on communities of practice, we

take the above one step further by proposing that these distinct community
of practice structural properties can be linked to performance. As
discussed above, within the field of communities of practice there is a
general inherent assumption that communities of practice have a positive
impact on organizational performance (c.f. Brown & Duguid, 1991). The
general assumption is that as members of a community of practice work
together, they improve their practice. Wenger (1998) argues that
communities of practice produce incremental improvements in work

practices yet they are not favorable to radical improvements. Similarly,
Lave & Wenger (1991) argues that communities are involved in
simultaneously producing both practical outcomes for customers as well as
learning for members. However, there are few empirical studies that
provide evidence of this positive relationship. This lack of empirical
research on community of practice performance is understandable since by
definition a community of practice is a fluid, emergent informal structure
(Brown & Duguid, 1991; Wenger, 1998). As a result, communities of
practice are extremely hard to pin down. In addition, there is no agreement
as to a performance measure that would capture the community of practice
as a whole, again due to their fluid nature (Brown & Duguid, 1991;
Wenger, 1998.). However, if we are able to develop the means to detect
and analyze communities of practice within organizations through the
development of structural properties, then we may be able to investigate the
relationship between these structural properties and performance. In
summary, through identifying and specifying structural properties of
communities of practice, we may then open the door for additional
theorizing on both the structural as well as the cognitive aspects of
communities of practice as well as for further empirical studies. Thus,
Research Purpose 1a is the following:
Research Purpose 1a: To conceptually develop the structural

properties of communities of practice and propose a series of
relationships between community of practice structural
properties and performance.
4.1.2 Research Purpose 1b: Structural Dimensions of Electronic

The second network of practice that we have chosen to investigate here is
that of an electronic network of practice. We extend our reasoning from
Research Purpose 1a that we may use social network analysis to help us
investigate the structural properties of communities of practice to electronic
networks of practice51. However, we find that we cannot merely apply the
51
We are not the first to apply social network analysis to online communities; however,
to date most of the applications of social network analysis to online communities have
focused on mapping the links that cross-posted messages establish between and not
118 CHAPTER FOUR
properties developed for communities of practice in Research Purpose 1a to

electronic networks of practice due to the different logics of these
networks.
To recapitulate our discussion in Chapter Two, electronic networks of

practice are similar to communities of practice in that they are a social
space where individuals working on similar tasks self-organize to help each
other and share perspectives about their occupational practice or common
interest. However, unlike communities of practice and the other kinds of
networks of practice where people know each other personally and form
dyadic, interpersonal relationships, participants in electronic networks of
practice are typically strangers, individuals form weak ties with the entire
network instead of with a select few, and interactions generally occur
through text-based, asynchronous, computer-mediated communication.
Moreover, the differences between an electronic network of practice and
the other networks of practice relate to the visibility of the interactions
between members. In electronic networks of practice such as those
characterized by listserv or bulletin board technologies, the posting of
messages is generally open. Thus, once posted, messages are visible to
everyone participating in the network. As a result, anyone searching for
advice can either post a new question, or in some networks search the
archived discussions to reuse knowledge that has already been exchanged
between other members.
As we have seen in previous chapters, it has been proposed that theories of

public goods and collective action can be modified and expanded to explain
knowledge exchange in electronic environments (Fulk, Flanagin, Kalman,
Monge, & Ryan, 1996; Monge, Fulk, Kalman, Flanagin, Parnassa, &
Rumsey, 1998). Thus, one way of investigating the structural properties of
an electronic network of practice is to apply theories of collective action
and public goods in our analysis. In the formal language of collective
action theory, we argue that the electronic network participants are the
within newsgroups nor do these newsgroups revolve around a work practice (e.g.,
Donath et al., 1999; Smith, 1999; Sack, 2000; Choi & Danowski, 2002). A cross-posted
message is a message that is posted to more than one newsgroup. The “Newsgroups”
header of the message displays the names of all the newsgroups to which the message
was posted. Reasons for posting to more than one newsgroup may include that the
message is relevant to more than one group or the author was unsure as to which
newsgroup to post.
interest group or the collective and the public good is the continuous stream
of knowledge.
To explain further, public goods are resources from which all individuals in
a collective may benefit regardless of whether they have contributed to
providing the good, such as a public park or public television (Kollock,
1998). Public goods have two specific characteristics that have
implications for their provision and use. First, a public good is a resource
that can be provided only if members of a collective contribute towards its
provision. It is non-excludable, i.e., the good cannot be withheld from any
member of the collective, even if he or she does not participate in the
production or maintenance of the good (Komorita & Parks, 1995). A
second characteristic is known as non-rival, meaning that the good is not
used up or depleted in its consumption, thus one person’s use of the good
does not diminish its availability to others in the collective (Shmanske,
1991). Public goods are generally considered to evidence both non-rivalry
and non-excludability. Since public goods are not used up in their
consumption due to non-rivalry, there is no incentive to add costs by
controlling access to the good through exclusion (Musgrave, 1959).
However, a connection between the two characteristics of non-rivalry and
non-excludability does not necessarily exist: a non-rival good can be
excludable while a non-excludable good can be either rival or non-rival
(Shmanske, 1991). Thus, true public goods are completely non-excludable
and non-rival; however, it is argued that many public goods exhibit these
characteristics to varying degrees (Kollock, 1998).
As mentioned above, we follow the suggestions of previous researchers and

propose that we view the knowledge of an electronic network of practice as
a public good. First, due to the open nature of an electronic network of
practice, the network’s knowledge is non-excludable. When participants
interact in the network, then all members may benefit from their discussion
and knowledge sharing, even though they did not contribute to its
production through either posting or responding. Second, the knowledge is
non-rival since when one person uses the knowledge gained in the network
of practice, it still remains available to the other network members.
When knowledge is treated as a public good, then all members participate

in supplying the knowledge to the network of practice, thus creating
knowledge flows. These flows generate new combinations of existing
120 CHAPTER FOUR
knowledge, which results in learning, innovation and new knowledge

creation. However, one of the main concepts in theories of collective
action and public goods is the social dilemma of the provision of public
goods. In other words, the optimal individual decision is to enjoy the
public good without contributing anything to its creation or maintenance.
In the context of electronic networks of practice, the rational decision by
participants of an electronic network of practice would be to free-ride or
lurk, merely reading the messages posted by others without contributing to
either asking questions or spending time engaging in the online discussions.
Yet if all members were to lurk, then there would be no public good
produced. This is the social dilemma. Thus, we are interested in
investigating the structural properties of an electronic network of practice
in order to help us understand the provision of the public good of
knowledge, given that individuals are better off not contributing and
instead free-riding on the efforts of others. As such, Research Purpose 1b
is the following:
Research Purpose 1b: To investigate the structural properties of

an electronic network of practice through the application of
theories of collective action and public goods.
4.2 Research Purpose 2: Performance and Networks of

Practice
In reviewing the network of practice literature, we find that with few

exceptions one of the general assumptions is that there is a positive
relationship between networks of practice and performance since networks
of practice are argued to be the nexus for the sharing and in some cases the
creation of valuable individual and group knowledge. Researchers and
practitioners alike have been increasingly advocating networks of practice
within recent years, and as a result, managers in numerous organizations
are attempting to support or even construct various forms of networks of
practice within and across their organizations (Wenger et al. 2002; Swan et
al. 2002). Similar to the broader set of organizational knowledge
management initiatives implemented to enhance performance through
knowledge sharing and creation, it seems that the hope of management is
that these efforts will positively affect individual behavior in the workplace
and thus ultimately drive increases in firm performance (Davenport &
Prusak, 1998; Wenger et al., 2002).
However, as revealed in the review of the empirical studies in Chapter

Three, there is a dearth of solid academic empirical support for this positive
relationship, with researchers paying little systematic attention to the
relationship to performance at any level. Several reasons for this can be
offered. One potential reason could stem from the assumption that learning
leads to improved performance. As discussed in Chapter Two, the original
research that resulted in the development of communities of practice was
based on situated learning theory in which individuals are argued to learn
through participating in a shared activity. Thus, the intuitive benefits of
networks of practice may seem to be obvious, thus requiring little support
for substantiation. Individuals mutually engage with others participating in
the same practice, learning to conduct their work-related tasks, and thus
improving their ability to perform their work-related tasks, with the
outcome being improved performance for both the individual, the network
of practice, and the organization. While this positive relationship to
performance may seem intuitive, as we discuss below, previous research
suggests, however, that there may be a negative relationship to
performance as well (Levitt & March, 1988; Leonard-Barton, 1992).
A second possible explanation for the dearth of empirical study focusing on

performance is that defining and measuring performance related to
networks of practice is extremely difficult. The ability to develop
appropriate performance constructs is hampered by the ethereal and
emergent nature of networks of practice; they are, by definition, extremely
hard to pin down. Any individual can potentially be involved in numerous
networks of practice, varying from an immediate community of practice to
a set of internet contacts on the other side of the world. Moreover, the
process of defining membership in certain networks of practice apparently
takes away their very essence because they thrive on their emergent nature.
One approach that could be used to address these difficulties is to bring the
network of practice thinking down to the level of the individual. Rather
than attempt to define the various networks of practice within and across a
firm’s boundaries, we propose that an individual’s performance at work is
associated with the extent to which he or she is a member of and
participates in different networks of practice. Thus, by measuring the
patterns of interaction of the individual with various networks of practice,
we suggest that we can predict to some degree his or her performance.
122 CHAPTER FOUR
In so doing, we contribute to the network of practice literature in an

additional way. One common trait of all networks of practice is that by
definition they are self-organizing. As a result, their existence and
sustainability are dependent upon individual members and their desire to
participate and share their knowledge within the network. However, as
mentioned above, the empirical studies to date by and large depart from a
specific type of network of practice with individuals already having made
the decision to participate in the particular network. Furthermore, with the
exception of the studies of electronic networks of practice, it is generally
assumed that individuals freely share their knowledge with other network
of practice members. However, previous research investigating knowledge
sharing activity implies that individuals make choices as to whether or not
to share their knowledge. In one article, it is even simply put, “To hoard or
to share? That is the question” (Boisot & Griffiths, 1999:664). Within this
line of questioning, considerable attention has been paid to why individuals
help each other and share their knowledge in face-to-face networks (e.g.,
Thibaut & Kelley, 1959; Schwartz, 1970) as well as in electronic networks
of practice (e.g., Constant et al., 1996; Wasko, 2002). Results indicate that
individuals base their decisions on potential costs and benefits to
knowledge sharing. For example, research by Allen (1977) provides
evidence that individuals consider the "cost" of an interaction with another
individual before engaging in any interaction. Additionally, as we saw in
previous chapters, motivations for knowledge sharing could be based on
rational, self-interest or collective interest, or both. For example, the
electronic network of practice research provides evidence that individuals
tend to share knowledge in the hope of increasing their reputation or
advancing the electronic community (Lakhani & von Hippel, 2000).
However, while studies have looked at the intangible antecedents to
participation within one particular network such as reputation and altruism,
researchers have paid scant attention to an individual’s participation in a
particular network and its effect on individual outcomes such as individual
performance. At the individual level, other than the categories of
community of practice participation proposed by Wenger (1998), there is
little discussion of this relationship.
In addition to this simply put question above, it is obviously not the only
question that needs to be asked when discussing knowledge sharing by
individuals. Clearly, if there is one general conclusion we can make from
the discussion and literature review presented in the previous chapters, it is
that individuals have a large variety of networks of practice within which

they may participate: internal vs. external, face-to-face vs. electronic, etc.
Previous research further supports the idea that individuals make choices
regarding their level of participation in various networks. For example,
Wellman & Gulia (1999) suggest that individuals operate with a portfolio
of relationships of various degrees of density according to the purpose in
belonging to the network. Additionally, although based on only 15 semi-
structured interviews of scientists in an inter-organizational consortium,
Andrews & Delahaye (2000) found that the respondents (1) deliberately
mediated the knowledge acquisition process by deciding from whom they
would seek potentially useful knowledge as well as from whom they were
willing to accept knowledge and (2) actively deciding with whom they
would share their knowledge52. As mentioned above, these choices are
primarily made based upon individual interests, which are diverse and
distributed across individuals. As a result, the process of knowledge
sharing among individuals is a rather fragile and uncertain activity (von
Krogh, 2002). For example, theories of self-interest propose that
individuals make what they believe to be rational choices to acquire
personal benefits through maximizing (or satisficing) their gains or
minimizing their losses (Monge & Contractor 1997). Thus, individuals
acting in this manner feel that there must be some reciprocal rewards for
participating in knowledge sharing, such as enjoyment, being challenged,
or improved task performance (Wasko & Faraj, 2000).
Research has also found that individuals are motivated by collective

interest, which includes “care” (von Krogh, 1998), prosocial behaviors
(e.g., “the right thing to do”) and an interest in advancing the collective
(Wasko & Faraj, 2000). In the context of networks of practice, then we
would argue that individuals make their own decisions as to with whom
they would like to interact and share knowledge (and thus in which
network of practice to participate) based on the potential returns from these
actions to themselves and/or to the network. Thus, individuals may expect
that participation in a certain network may result in certain outcomes.
While several studies have investigated individual interests in terms of the
antecedents to choices to participate within a particular network of practice,
(e.g., Lakhani & von Hippel, 2000; Wasko & Faraj, 2000), researchers
52
These authors propose a psychosocial filter involving social confidence, perceived
credibility, and perceived trustworthiness as a means of understanding the antecedents
to these decisions.
124 CHAPTER FOUR
have paid scant attention to the relationship between an individual’s

participation and knowledge sharing in a particular network and individual
outcomes such as individual performance. Thus, additional questions to
ask are “In which networks do individuals participate and share their
knowledge? and “Does the level of participation in various networks of
practice result in different outcomes?”
Tying this back to our above discussion regarding performance, we may

then fill these research gaps by specifically focusing on the individual and
the relationship between the degree of participation in various networks of
practice and his or her individual performance. Thus, our second research
purpose becomes the following:
Research Purpose 2: To investigate the relationship between

individual participation in various types of networks of practice
and individual performance.
In order to fulfill this research purpose, we conduct several studies in which

we develop a set of hypotheses relating the participation and knowledge
sharing of individuals in various networks of practice and implications for
individual performance. Since the focus of this thesis is on networks of
practice that are related to work practice within and across business firms,
we turn to the knowledge-based view of the firm literature to further help
develop our hypotheses53.
The knowledge-based view of the firm (KBV) has developed in response to

recent advances in strategic management thought that suggest that
knowledge is the most strategically significant resource of the firm (Grant,
1996a,b) and that sustained competitive advantage and superior corporate
performance are determined by heterogeneous knowledge bases and
capabilities among firms (Eisenhardt & Santos, 2002). Interestingly and
not too surprisingly, KBV researchers have yet to reach a consensus in
several areas. For example, on the one hand, several proponents such as
Grant (1996a,b) view knowledge as a resource while others such as
Spender (1996) argue that the firm should be seen as a system of knowing
activity rather than a system of applied knowledge bundles that can be
moved around the organization (Eisenhardt & Santos, 2002). To date,
53
For a review and discussion of the knowledge-based view of the firm, see Eisenhardt
& Santos (2002) or Kaplan, Schenkel, von Krogh, & Weber (2002).
empirical research on the knowledge-based view tends to fall under the

former perspective with research focusing on four categories of specific
knowledge processes: knowledge sourcing, internal knowledge transfer,
external knowledge transfer, and knowledge integration (ibid). The fourth
category, knowledge integration, stems from Grant (1996a) who argues that
the firm’s primary task is to integrate the specialized knowledge of its
members54 and that competitive advantage results from how effective firms
are in performing knowledge integration. As such, much of the research on
knowledge integration takes a micro view of interaction, primarily within
organizations, and focuses on investigating how the specialized knowledge
of individuals is integrated in firms. For example, Eisenhardt (1989b)
conducted an inductive study of how top management teams in the
computer industry integrated their different functional and personal
perspectives to achieve strategic decisions.
Since we also take a micro view of knowledge processes in this thesis, it

follows then that we use this lens of knowledge integration in our empirical
studies. In his knowledge integration argument, Grant (1996a) posits that
the effectiveness of knowledge integration depends upon the efficiency, the
scope, and the flexibility of knowledge integration. Efficiency refers to
how well the specialized knowledge of a firm’s individuals is integrated
and is dependent on a common language of discourse developed thorough
frequent interactions between individuals such as that which occurs in
communities of practice. The scope of knowledge integration refers to the
different types of specialized knowledge being integrated – the more
complex the scope, the greater the difficulty for competitors to replicate.
Flexibility of integration reflects extending existing capabilities through
boundary spanning activities in order to access and reconfigure additional
knowledge through both internal and external integration. Sustaining a
competitive advantage requires flexibility and the creation of new
capabilities by bringing in new knowledge and reconfiguring existing
knowledge. Intra-organizational networks of practice promote the internal
flexible integration of knowledge across a firm’s internal boundaries while
54
This is, of course, consistent with other established theories of the firm. Barnard
(1938) spoke in terms of conscious, deliberate, and purposeful cooperative action
between individuals as the reason for the existence of the firm. He further argued that
organizations are cooperative systems that serve to integrate the contributions of
individuals. Transaction Cost Economics would also argue that cooperative action can
be achieved most efficiently within the firm when the complexities of specifying the
contracts between the cooperating parties become too great.
126 CHAPTER FOUR
inter-organizational networks of practice provide the means to access

knowledge outside of the firm’s boundaries in order to enhance external
integrative flexibility.
However, this theory of knowledge integration represents a paradox: a

focus on the efficiency of integration may hinder flexibility and the ability
to create new knowledge and innovations. For example, although it has
become widely recognized that participation in communities of practice
supports efficient knowledge exchange, learning and incremental
innovation, tightly knit communities of practice may lead to the “not
invented here” syndrome or the resistance to new ideas not locally
developed. In addition, the knowledge in a tightly knit community of
practice may be largely redundant. For example, Granovetter (1973, 1983)
argues that closely-knit clusters in which individuals are well-acquainted
and interact often are characterized by knowledge that is redundant.
Knowledge is likely to be quickly shared and commonly known, thus
individuals receiving knowledge from someone within the cluster may well
already have received this knowledge from someone else. However, weak
ties, i.e., characterized by a relatively low involvement of time, emotional
intensity, intimacy, and reciprocity, are instrumental to the diffusion of new
knowledge. Weak ties provide access to knowledge from people who
travel in different circles or engage in different activities. The knowledge
they carry is more likely to be novel and not otherwise available. Thus,
community of practice interactions may provide little additional knowledge
over what an individual may already know, thus impeding the ability to
develop new and creative ideas (Granovetter, 1973, 1983). As a result,
communities of practice may evolve into core rigidities and competency
traps – inappropriate knowledge sets that preserve the status quo and limit
new insights, resulting in gaps between the knowledge of the firm and
changing market conditions (Levitt & March, 1988; Leonard-Barton,
1992). However, following weak tie theory, knowledge available in
distributed networks of practice is more likely to be less redundant due to
the weaker nature of the ties in these networks and thus may facilitate
flexible knowledge integration.
In addition, Grant’s theory focuses primarily on issues of coordination,

without referring to issues of “cooperation”. This theory then leaves out a
key component by assuming that people are willing to share knowledge
openly and freely if provided with the structures and opportunities to
interact. However, as seen above, prior research suggests that individuals

do not give away their help and advice to others for free (e.g., von Hippel,
1987; Schrader, 1991). Rather, they tend to participate in reciprocal
behaviors involving the trading or exchange of knowledge. Furthermore,
previous research on boundary spanning activities across a firm’s legal
boundaries has indicated that this reciprocal behavior may involve
“leakage”, or the flow of company proprietary knowledge across firm
boundaries (Mansfield, 1985; Von Hippel, 1987; Schrader, 1991). Thus,
participation in knowledge trading and the resulting potential for informal
proprietary knowledge flows across an organization’s legal boundaries are
of particular strategic interest to management since such activity may
impact a firm’s competitive advantage. Yet, it is very difficult for firms to
manage and evaluate the benefits since it occurs “off the books” with
employees generally acting completely on their own with no managerial
influence and no documentation of the trade (von Hippel & Schrader,
1996).
In summary, key strategic issues for organizations interested in

successfully managing their knowledge resources involve understanding in
which networks of practice individuals participate, whether internally or
across a firm’s legal boundaries and how they access knowledge within
these different networks as well as how to balance efficient knowledge
integration with demands for flexible integration. Investigating
participation and knowledge sharing in various networks of practice should
help shed some light on these issues. As a step in this direction and as
stated above, our approach then is to bring the level of analysis down to the
individual level as opposed to focusing on the organizational level by
investigating the relationships between the participation and knowledge
exchange of individuals in various networks of practice and individual
performance. It is important to be clear here that our research focus is not
on the antecedents of these choices, such as the how or why individuals
make certain choices about participation. Rather our interest is focused on
the relationship between an individual’s participation in various networks
of practice once the choice is made and individual performance. Thus, in
addition to contributing to the literature on networks of practice by filling
the research gap of the relationship between individual performance and
network of practice participation, we also hope to contribute to the
knowledge-based view of the firm.
128 CHAPTER FOUR
4.3 Summary of Research Purposes
In table 4.1 we present the two research purposes and how they correspond
to the seven empirically based articles in Appendix Two. In addition, we
present the research purposes and their corresponding articles in the matrix
of networks of practice in order to show how they fit into the overall
research on networks of practice (figure 4.1). In the next chapter, we will
proceed to present the methods and empirical data collections used to fulfill
our two research purposes.
Table 4.1 Overview of Research Purposes and Corresponding Articles

Research Purpose Article
RP1a: To conceptually develop the structural properties of Article 1
communities of practice and propose a series of relationships
between community of practice structural properties and
performance
RP1b: To investigate the structural properties of an electronic Article 2
network of practice through the application of theories of
collective action and public goods
RP2: To investigate the relationship between individual Articles 3-7

participation in various types of networks of practice and
individual outcomes
Figure 4.1 Positioning of Research Purposes in Network of Practice

Matrix

Research
Co-located
Purposes 1a, 2
(Articles 1, 0/ 0/
3,4,5,6)
Non-co-located
Research Research
0/ Purpose 2
(Articles
Purpose 2
(Articles 3,4,5,6)
3,4,5,6,7)
organizational
Research Research
Inter-
0/ Purpose 2
(Articles 3,5,6)
Purposes 1b, 2
(Articles 2,
3,5,6)
CHAPTER FIVE
Research Methodology
THIS CHAPTER PRESENTS the research methodology employed in the

seven empirical studies. For the purposes of this thesis, we conducted a
series of cross-sectional studies using a variety of techniques: surveys,
interviews, text analysis, social network analysis, etc. All the studies
presented here were conducted in field settings and the empirical basis of
this thesis consists of four different research sites: an international
contractor consortium in the construction industry in Scandinavia, two
multinational consulting firms, and one inter-organizational electronic
network of practice whose members are lawyers dispersed across the
United States. In order to maximize the relevance of the studies and to
enhance the external validity of the findings, research sites were chosen for
two reasons: 1) an emphasis on knowledge work and 2) knowledge as the
most valuable asset available to these individuals. As mentioned above, the
work of the individuals in all of these studies primarily consists of
processing, articulating, applying and disseminating knowledge (Wasko,
2001) as opposed to making tangible objects with their hands. Six of the
articles presented here are primarily based on quantitative data collected in
the form of surveys (from paper to web-based), and one is based on
qualitative data gathered in semi-structured interviews. As mentioned, all
the studies were conducted using a cross-sectional design. The group of
quantitative studies will be discussed first followed by a description of the
qualitative study. The chapter concludes with table 5.1 providing an
overview of the various research sites and methods used.
5.1 Quantitative Studies
For each of the quantitative studies, an initial qualitative study was

conducted that comprised a set of face-to-face interviews in order to
130 CHAPTER FIVE
examine the concepts to be used prior to the construction of the survey

instrument. In constructing the questionnaires, reviews of previous studies
were conducted in order to find existing scales. Where possible, these
scales were used or adapted to our purposes. As a result, the questionnaires
were based upon a mixture of established scales from the literature and our
own measures of constructs relevant for our purposes. Most questions had
pre-defined response alternatives based on seven-point Likert scales.
Where appropriate, the independent frequency variables were transformed
to convert responses from an interval scale to a ratio scale. Moreover,
background variables important for each study were collected, e.g., age and
experience of respondents. Pilot tests of all instruments were conducted
with modifications made to the surveys based on the results of the pilot
tests. The format for each questionnaire was consistent across all
participants within each study to reduce confounds due to the survey
medium. Throughout the data collection processes, individuals were
assured that their responses would be kept confidential and that all results
would be presented only on an aggregate level. In one of the studies, there
was an incentive provided for participation (the second Icon study, Article
6).
5.1.1 Internal Reliability and Validity

When using measuring instruments, there are two important questions to
answer according to Kerlinger (1986): (1) what is the reliability of the
measuring instrument and (2) what is the validity of the measuring
instrument. Thus, a crucial step prior to testing any theoretical model is the
assessment of the accuracy of the measurement model. The goals of
assessing the accuracy of the measurement model are to demonstrate that
the measures used are valid and that they adequately reflect the underlying
theoretical constructs.
Internal Reliability. Research instruments that have a high degree of

internal reliability consistently measure what they are intended to measure
because they are dependable, stable, consistent, and predictable (Kerlinger,
1986). The degree of precision and the accuracy determine an instrument’s
reliability and a reliable one is minimally affected by random measurement
error. Thus, if the instrument is repeated, it will yield similar results.
When constructs are used, an assessment of reliability will indicate how
accurate on average a construct created by adding items together will
RESEARCH METHODOLOGY 131
measure the “true” score of a population55. The internal consistency

method is often used to determine the reliability of empirical
measurements. Individual survey items that make up a theoretical
construct are assessed for inter-item reliability, and this evaluation requires
multiple items for each construct. In most cases, we used Cronbach’s
coefficient Alpha to measure inter-item reliability. Acceptable values of
Cronbach’s Alpha for perceptual measures should exceed 0.7 (Nunnally &
Bernstein, 1994). Values less than 0.7 imply that the items underlying the
construct may be unrelated, or may be measuring more than one construct.
With the exception of the first exploratory study at Icon, the alphas (or their
equivalent, internal composite reliability, ICR) were above 0.7.
Internal Validity. In addition to inter-item reliability, it is also necessary to

assess that the items underlying constructs demonstrate internal validity.
One method for assessing convergent and discriminant validity is factor
analysis. Factor analysis can be used to determine whether items
underlying a theoretical construct have a high degree of correlation among
one another compared to their correlation with items underlying other
constructs. However, scholars have cautioned the use of this method since
naïve and simplistic interpretations of factor analyses may be misleading
rather than validating (Carmines & Zeller, 1979). Taking these words into
consideration, factor analysis was conducted where appropriate. Results of
these analyses showed clean factor separation in most cases. In instances
where items did not exhibit adequate loading, these items were dropped
from the measurement model and the internal reliability was recalculated.
Multi-item constructs were then calculated by taking the average of the
items.
For each study below, the sites as well as the methods used are described.
However, in order to avoid repetition with the studies, these are not
described in great detail here, rather the reader is referred to the relevant
study in Appendix Two.
55
See Nunnally & Bernstein (1994) for a more elaborate and scholarly discussion of
reliability assessment.
132 CHAPTER FIVE
5.2 Research Methodology for Research Purpose 1
While the first overarching research purpose is to describe the structures of

networks of practice through the application of social network analysis, this
purpose is divided into two sub-purposes. Since these sub-purposes focus
on two different networks of practice (a community of practice and an
electronic network of practice), we collected two different sets of data.
5.2.1 Research Study at Sundlink AB - Article 1

To refresh our memory, Research Purpose 1a is the following:
Research Purpose 1a: To conceptually develop the structural

properties of communities of practice and propose a series of
relationships between community of practice structural
properties and performance
This study is presented in Article 1 and uses data collected from Sundlink
Contractors, an international contractor consortium that designed and
constructed the Öresund Bridge, a five-mile multi-level bridge connecting
Denmark and Sweden, during 1996-2000. The choice of this site presents
an interesting opportunity to examine communities of practice based on
two factors: a continuous stream of emergent problem situations and the
ability to define joint enterprise. The continuous stream of emergent
problem situations resulted due to the nature of the project: a highly
complex infrastructure project of immense size, stringent quality
requirements, well-defined completion time, and harsh environmental
conditions.
In terms of defining joint enterprise, Sundlink Contractors utilized a formal

quality system based upon ISO 9000, which articulated the work processes
and procedures. It is within this quality system that we have defined the
joint enterprise for communities of practice in this study (Schenkel, 2002).
In particular, we look at the management of ”deviations” or situations in
which articulated procedures or processes are not followed or articulated
objectives are not achieved. The management of deviations requires 1) the
use of already existing work methods, 2) a change in existing work
methods, or 3) the development of new technical solutions. Thus,
examining deviations provides the context for exploring joint enterprise
(management of the deviation within the project), a shared repertoire (the

common means/behavior in which the incidents are approached and
managed), and mutual engagement (the collaboration of multiple
individuals/groups of different and/or the same competencies).
Data were collected through a questionnaire administered during a nine-

month period ending in May 1999 as part of a doctoral project (Schenkel,
2002). The data collection occurred mid-way through the project. This
time point was chosen in view of evidence that communities of practice do
not form at once, but more gradually, based upon recurrent events
(Wenger, 1998). Earlier data collection might have preceded the formation
of communities of practice. Two types of data were collected: 1)
communication patterns in managing deviations, e.g., whom the respondent
contacted both within and outside the organization for advice in situations
which deviated from prescribed ISO 9000 standards, and 2) socio-
demographic information such as age, education, and experience data - all
potential factors that can influence the formation and maintenance of
communities of practice. The population was delimited by choosing those
who were not construction workers, i.e., those who had a managerial or
support function. These individuals were excluded from this study in view
of their limited role in the actual management of deviation. In total, 137
people of both an operational as well as support character were included in
the population and 120 people responded to the questionnaire, resulting in
an 87.6% response rate.
For the purposes of this article, we perform three separate analyses to

determine to what degree the unit of analysis fulfills the structural
properties of a community of practice to illustrate the structural properties
of communities of practice. The three units of analysis are 1) the overall
project based on relationships between individual project members
regardless of department membership, 2) each department based on
relationships between the department’s individual members, and 3) the
overall project based on the aggregated individual relationships between
departments. We analyzed the data using SPSS and the UCINET network
analysis software package (Borgatti, Everett, & Freeman, 1999) and
importing it into Krackplot (Krackhardt, Blythe, & McGrath, 1994), a
program used for the graphical analysis of networks.
134 CHAPTER FIVE
5.2.2 Research Study of a US Professional Legal Association - Article 2

Research Purpose 1b is the following:
Research Purpose 1b: To investigate the structural properties of

an electronic network of practice through the application of
theories of collective action and public goods.
This study is found in Article 2 and data were collected from a single inter-
organizational network of practice of lawyers in a US professional legal
association. All association members have access to an electronic network
of practice as part of their membership benefits, yet participation is
voluntary. This electronic network of practice is supported by “bulletin
board” technology, similar to that of Usenet newsgroups where questions
and responses are connected in a “thread”, resembling a conversation.
During the two months of April and May 2001, there were 2,460 messages
posted to the network by 526 unique individuals. (The name of the person
posting was included in each message.) Individuals were then chosen to
take part in this study based on their electronic network of practice
participation, which consisted of posting a message to the network during
the two months under investigation. Each participant was then sent an MS
Word questionnaire distributed as an email attachment, and 152 valid
responses were received for a response rate of 29%. These data were
collected as part of a doctoral project (Wasko, 2002).
For this article, we examined all bulletin board messages to determine the
identity of the person posting, and the messages were then coded as seeds
(the first message in a thread), singletons (seeds without responses),
questions, responses, or other. We then built a social network matrix
consisting of all 526 participants to determine who was responding to
whom, creating a directional, social tie. Building upon Wenger’s
categories and based upon the analysis of messages, we created four
categories of participants: outsiders (people who posted seeds, but never
received a response), seekers (people who posted only questions),
periphery (people who posted 10 or less responses) and insiders (people
who posted more than 10 responses). Using UCINET software (Borgatti et
al., 1999), we analyzed the ego network of each individual to determine
centrality in terms of “in degree”, i.e., the number of times other people
respond to an individual, and “out degree”, i.e., the number of times an
individual responds to others. Similar to the previous study, we imported

the data into Krackplot (Krackhardt et al., 1994) to create a graphical
illustration of the network. We then used both the objectively collected
message postings as well as survey results to perform our analysis.
5.3 Research Methodology for Research Purpose 2
Restating Research Purpose 2, it is to investigate the relationship between

individual performance. The five articles, Articles Three to Seven, that
address this research purpose resulted from research conducted at two
multinational consulting firms, Icon Medialab and Cap Gemini, and three
high-technology multinationals: Hewlett-Packard plus two others who wish
to remain anonymous. Data were specifically collected for the purposes of
this thesis. Before presenting a description of the different research sites
and methods, we discuss two methodological issues common to all the
articles: operationalization of network of practice participation and
measurement of individual performance.
Operationalization of Network of Practice Participation. First, in terms of

operationalization, few researchers have attempted to understand the
relationship between networks of practice and performance. This is
understandable because networks of practice are – by definition extremely
hard to pin down. Any individual can potentially be involved in numerous
networks of practice, varying from one’s immediate workgroup to a set of
internet contacts on the other side of the world. Moreover, the process of
defining the membership of networks of practice apparently takes away
their very essence because they thrive on their emergent nature. Bearing
this in mind, it is not surprising that the vast majority of studies specifically
focused on networks of practice are ethnographies and qualitative studies
rather than quantitative studies using survey instruments. Thus, there are
significant methodological challenges in studying communities of practice.
The primary concern is that the concept is typically defined in such a way
that all emergent interactions, inside or outside the firm, could represent
participation in networks of practice. If this broad definition is accepted,
then the concept becomes very difficult to research in a rigorous manner
because nothing can be excluded. The theory, in other words, cannot be
falsified.
136 CHAPTER FIVE
However, the approach in these articles is to bring the network of practice

thinking down to the level of the individual. Rather than attempt to define
the various networks of practice within and across the firm’s boundaries,
the logic here is that individuals are able to draw from their networks of
practice to solve problems they encounter in the course of their work, and
that they also contribute back to these networks of practice in a reciprocal
manner. By examining the ways in which an individual acquires
knowledge to address work-related problems, we will see that some
knowledge is gained through access to “codified” sources such as internet
websites or company databases, but most is gained through interaction with
other people in the firm and outside. The premise, in other words, is that
the frequency and quality of the interaction an individual has with specific
groups of individuals is a manifestation of the networks of practice in
which he or she participates. And the extent to which an individual
actively participates in the various networks of practice will ceteris paribus
be associated with an individual’s performance at work. Thus, by
measuring the patterns of interaction and knowledge exchange of the
individual with various networks of practice and through various media, we
argue that we can predict to some degree his or her performance. Thus,
while two levels of analysis are of interest in these studies: the individual
and the network of practice, the primary unit of analysis, the individual, is
appropriately based on the theoretical development of the variables and the
proposed causal relationships (Klein, Dansereau, & Hall, 1994).
Measuring interaction between individuals is a multidimensional

phenomenon that can be conceptualized and measured across a number of
attributes, such as frequency, mode, openness, density, directionality, and
so on (Allen, 1977; Tushman, 1977; Jablin, 1979; Gupta & Govindarajan,
1991). In the studies conducted as part of this thesis, we operationalize and
measure interaction and knowledge exchange in various types of networks
of practice primarily in terms of frequency, with the exception of the
second Icon study (Article 6) in which we also capture directionality.
However, while we do recognize that this particular operationalization does
not adequately capture either the content or the quality of the interaction
and knowledge exchange, the notion of frequency of interaction as an
indicator of the intensity of the tie between two persons has a long
academic tradition (cf. Homans, 1950; Granovetter, 1973), and influential
empirical literature (e.g., Allen, 1977; Tushman, 1977) has made effective
use of the frequency measure in different settings (Ghoshal, Korine, &
Szulanski, 1994). We follow these studies as well as more recent ones

(e.g., Ghoshal et al., 1994) and adopt frequency as our measure of an
individual’s interaction with other network of practice members. However,
in so doing, we acknowledge that we limit ourselves in terms of both the
theoretical and normative implications of our findings.
Measurement of Individual Performance. Regarding the second

methodological issue, as mentioned above, individual performance is “the
most widely studied criterion variable in the organizational behavior and
human resource management literatures” (Bommer, Johnson, Rich,
Podsakoff, & MacKenzie, 1995: 587). However, measuring performance
has proven to be a difficult task. The search for truly reliable,
uncontaminated, objective indicators of individual accomplishments within
organizations has proven unsuccessful (Campbell et al., 1992), thus,
researchers continue to debate the best means by which to measure
performance.
Current measurement approaches within organizational behavior and

human resource management literature include subjective measures (e.g.,
self, peer, and supervisor ratings) and “objective” measures that are based
on direct measures of countable behaviors or outcomes (e.g., total sales
volumes or sales commissions for salespeople). Research has provided
evidence that subjective and objective measures are not interchangeable
since even the “best” subjective ratings have been found to correlate only
to a low degree with “objective” measures (Bommer et al., 1995; Rich,
Bommer, MacKenzie, Podsakoff, & Johnson, 1999). Furthermore, in this
research, since we have defined individual performance as an individual’s
actions and not the concrete outcomes of an individual’s actions, we are
primarily interested in subjective measures. However, researchers are more
or less in agreement that perfectly reliable and valid third party
performance ratings are unattainable since they are subject to a variety of
biases, such as external conditions, the experience of the rater with the job
being evaluated, or the ability of the rater to observe the ratee (Borman,
1978; Weekley & Gier, 1989). Studies have shown that the correlations
between the various subjective measures tend to be less than “perfect” (see
Bommer et al. (1995) and Harris & Schaubroeck (1988) for a discussion).
For example, in one meta-analysis, Harris & Schaubroeck (1988) found a
correlation of .35 between self-ratings and supervisor ratings.
138 CHAPTER FIVE
As for performance measured by self-ratings, a number of previous studies

on managers and professionals have found self-reporting measures to be
superior to third party measurements (e.g., Heneman, 1974; Wexley,
Alexander, Greenawalt, & Couch, 1980). For example, Heneman (1974)
found self-rating measures to have less leniency, restriction of range, and
halo error than ratings by superiors. Furthermore, a review of the literature
on self-rated performance found strong support for their use (Busch &
Bush, 1974). However, other researchers argue that self-rated measures are
inflated due to egocentric bias (Churchill, Ford, Hartley, & Walker, 1985;
Harris & Schaubroeck, 1988). Thus, in summary, there exists no one
“best” measure of individual performance.
In each of our studies, we discussed with management which performance

measures were possible to collect. In our first study (Article 3-Icon Study
1), we were able to collect both self-reported ratings as well as ratings by
the respondents’ immediate supervisor. To establish the extent of inter-
rater reliability, i.e., the level of agreement on performance between the
individual and his/her immediate superior, we used Cohen’s Kappa, one of
the most widely used measures of inter-rater reliability, which is calculated
as follows (see Perreault & Leigh, 1989).
K = (F0 – Fc) / (N – Fc)
In this equation, N is the total number of judgments made by each judge, F0

is the number of judgments on which the judges agree, and Fc is the number
of judgments for which agreement is expected by chance. We assumed that
if the individual and his/her boss rated performance to within one point
they agreed (e.g., one person circles 2, the other circles 3), which means
that Fc, the number of agreements by chance, is 38%. On this basis the
Cohen’s Kappa scores for the two measures of performance examined were
0.55 and 0.63 respectively. Following this analysis, we interviewed a
number of individuals about the performance measures. It became clear
that in many cases (especially administration and sales), the level of
supervision was rather low, and as a result the supervisor or project
manager in question was typically not well informed about the individual’s
performance. However, our results seem to indicate a higher correlation
than that generally found in organizations. As discussed above, a meta-
analysis across a number of professions found only a moderate correlation
of .35 between self and supervisor ratings. The self-supervisor rating
correlation for professional/managerial jobs has been found to be even

lower than for that in blue collar/service jobs. One explanation for this
lower correlation is that professional/ managerial jobs can be argued to be
more ambiguous than well-defined blue collar/service jobs (Harris &
Schaubroeck, 1988). Thus, since the jobs at Icon fall into the category of
professional/managerial jobs, it could be argued that supervisors and
individuals at Icon agreed on individual performance to a relatively higher
degree. Furthermore, at Cap Gemini, through discussions with
management, it became apparent that supervisor ratings or peer ratings
would be difficult to obtain due to internal policy. Accordingly, we opted
to measure individual performance via self-reporting measures in our
studies both at Icon (Articles 3 and 6) and at Cap Gemini (Articles 4 and
5). This choice also facilitated our ability to make comparisons across the
studies. Having discussed some of the methodological challenges present
in Research Purpose 2, we now turn to the description of the research sites
and methods for the respective articles. We begin with Cap Gemini.
5.3.1 Research Studies at Cap Gemini – Articles 4 and 5

The research for Articles 4 and 5 was undertaken in the Nordic operations
(Denmark, Finland, Norway, and Sweden) of Cap Gemini and was
performed prior to the merger of Cap Gemini and Ernst & Young
Consulting. As a result, the company description considers only the Cap
Gemini organization. At the time, Cap Gemini was Europe’s largest IT
services and management consulting company with more than 40 offices
and 4,500 employees in the Nordic region alone.
Within the Nordic region, Cap Gemini had numerous networks designed to
enhance the company’s knowledge management activities. We chose
participants in one electronic network of practice, the NCN MS, because it
was recognized as a successful, vital conduit of knowledge exchange. This
electronic network had 345 members spread across the Nordic countries
and the members of this network all worked with applying Microsoft
products in their responsibilities with Cap Gemini. This particular
population was chosen for the study to ensure that research subjects had
access to internal and external sources of information and know-how and
had familiarity with using communication technologies that supported
information and know-how exchange in electronic networks. In addition,
the job responsibilities of the members of the NCN MS electronic network
140 CHAPTER FIVE
required a considerable amount of creativity as new problem situations

constantly arose due to the rapid pace of change in information technology
as well as the diversity among client project demands. This helped ensure
that the population chosen for this study had to balance both general job
performance and demands for creativity.
Data collection was conducted as an email attachment during January 2000.

Previous research has suggested that electronic surveys using scale-type
questions are no less valid than paper surveys (Liefeld, 1988). It has also
been found that some subjects prefer electronic surveys to paper (Newsted,
1985), and that email responses may even be more valid (Kiesler &
Sproull, 1986). We received a total of 83 usable survey responses from the
345 participants with valid email addresses for a response rate of 24%.
This “low” response rate could be argued to be higher since research has
shown that mailing lists contain a very high percentage of individuals who
are members of the electronic network of practice but do not participate
(Nonnecke & Preece, 2000). In Article 4, we use correlation analysis to
investigate our research questions. In Article 5, we test hypotheses using
partial least squares (PLS), and perform two separate analyses
independently for each dependent variable (creativity and general
performance).
5.3.2 Research Study at Icon Medialab – Articles 3 and 6

The choice of Icon Medialab (Icon) as the second site was motivated due to
several reasons. First, in order to conduct a study of individuals’
participation in various networks of practice, including the recently
emerging electronic networks of practice, it was necessary to find a firm
where a large proportion of the employees were working on a day-to-day
basis with the latest internet communication technology. Icon is an internet
consulting firm, and as such employees in all functions are not only
extremely adept at using new internet-based communication media such as
bulletin boards, chatrooms, email, etc. but they also use these to a high
degree in their everyday work. A second reason for choosing Icon is that,
at the time of this study, it was a medium-sized multinational with offices
in several countries and as such had emergent intra-organizational networks
of practice. A third reason is that the company encompasses a wide variety
of functional competencies, e.g., system architecture, programming,
management consulting, art direction, project management, human
computer interface, etc, thus providing the possibility to both compare

across functional competencies as well as generalize across these functions.
It should be noted that Icon’s largest office was based in Stockholm during
the time of the data collection. This was not only convenient, but
Stockholm is also quite an opportune location for studying such a firm
because Sweden is at the forefront of digital communications technology.
The country has one of the highest penetration rates in the world of mobile
telephones and internet subscriptions per capita, and Stockholm is a
recognized high-technology “cluster”. Icon was one of many start-up
Internet firms in the area (founded 1996), and it was selected as one of the
world’s best 350 small companies in 1998 by Forbes (Forbes, 1998).
Additionally, Icon is one of the few internet consulting firms established
under the IT boom that is still in existence today.
We conducted research at Icon Medialab at two different times since it is

the basis of two different studies. The first study took place in the fall of
1998 when the firm comprised 242 employees, and the second study
occurred in 2001 when the firm employed 1698 individuals. As evidenced
by the difference in the number of employees, Icon experienced an
extremely high growth rate during the years between the two studies.
Growth in the firm during these years was through greenfield operations as
well as acquisition and mergers.
Icon Medialab was founded in March 1996 in response to the rapid growth
of the internet. The company’s mission is to facilitate the creation of
competitive advantage for its customers through the incorporation of the
internet in customer operations. Products and services include internet
websites, intranets, extranets, and e-commerce solutions. Icon Medialab’s
clients ranged from the Swedish Postal Service and Compaq to British
Petroleum and Volkswagen. By 2001, the company had offices in Europe,
the United States, and Asia. At the time of the first study, the company had
242 employees with 46% of these in Sweden. The remaining employees
were spread throughout offices of 10-25 employees in Spain, USA,
Finland, Denmark, Germany, Belgium, England, and USA. During the
second study, the company had 1698 employees in 28 offices across Asia,
the US, and Europe.
142 CHAPTER FIVE
A strategy of rapid global growth was developed by the founders at the

company’s inception. One of the means by which Icon hoped to achieve
profitable growth was through the reuse of knowledge developed
throughout its different projects. In fact, management set a target that more
than 50% of all projects should include already proven successful products
or services. Thus, Icon Medialab invested heavily in building its structural
capital, with the key objectives being to transfer and reuse knowledge
complemented with follow-up and reporting.
In addition, Icon Medialab is unique in its representation of a mixture of

competencies under the same organizational umbrella. These disciplines
include Technology, Design, Usability Engineering, Statistics and
Analysis, Media and Entertainment, and Business Strategy, representing
the six sides of the “Icon Cube”. Thus, Icon Medialab brings together art
directors, behavioral scientists, copywriters, journalists, scriptwriters,
animators, TV-producers, software programmers, management consultants
and web designers, with accounting, personnel, and administration
completing the organization.
5.3.2.1 First Icon Study – Article 3

For the first study, we conducted two phases of data collection and
analysis. The first phase was conducted in the Stockholm office, in which
thirty in-depth field interviews at different levels of the company were held
from May 1998 to June 1998. Extensive written material was also
collected from the company. The second phase of the data collection
during the fall of 1998 involved a postal questionnaire sent to all 242
employees of Icon Medialab at their local offices. Of the 242
questionnaires, 203 usable questionnaires were collected, an 84% response
rate. In addition to these individual questionnaires, each of the managing
directors of the eight subsidiaries and seven managers at the Stockholm
office were asked to complete a questionnaire relating to the performance
of the individuals at their office.
We then tested the propositions developed in Article 3 through a series of

stepwise regression models. We chose the stepwise approach primarily
because of the small sample size and the relatively large number of
independent variables. Also, the exploratory nature of the study makes it
appropriate to work with a rather larger number of independent variables

than would normally be the case.
5.3.2.2 Second Icon Study – Article 6

In order to further address Research Purpose 2, one firm in which we could
conduct a social network study designed to investigate the patterns of
interactions and participation in networks of practice by all individuals
within the firm had to be found. This decision to investigate only one firm
is in line with the social network field since the investigation of only one
firm is common in network studies (see Marsden, 1990; Hansen, 1996).
In order to conduct social network analysis, a response rate of at least 80%

of the individuals in the entire network is required since holes in the
network caused by non-respondents can easily distort the results.
Additionally, social network data tend to be more challenging to collect
than ordinary survey data. It is difficult for social network surveys to be
anonymous since the method generally requires identification of each
individual. This often presents problems since it is quite common that
individuals feel that this type of survey is a breach of their privacy, posing
questions about an individual’s personal connections such as “With whom
do you eat lunch?” Thus, it is not surprising that the majority of
sociocentric network studies on individuals tend to be at smaller
organizations or within organizational divisions or use some form of
publicly available data.
As a result, we needed to find one firm where a high level of access and
support could be provided by the company’s management. Through the
first study at Icon, a good rapport was developed with the company’s
management, and after consultation with senior management, we were
provided access for this study. Since this study is one of the few of its
kind, a lengthier description of the data collection is included here.
Although the first study at Icon had been conducted three years prior, it
was necessary to conduct a new set of interviews since the focus of the
second study included the social network component. Thirty-five
interviews were conducted throughout the firm to gain an understanding of
the various networks of practice within the firm as well as the different
inter-organizational ones in which Icon individuals participated. We
144 CHAPTER FIVE
created the survey in English since this is the official company language
and designed it as a web-based questionnaire for all employees of Icon
Medialab at their local offices. It is important to note here that all
employees have access to their own computer and the internet since the
majority of the work of all employees across task groups and hierarchical
levels is performed using the computer. In addition, management placed
few constraints on employees regarding the internal or external use of any
form of computer-mediated or other communication channels.
We created this questionnaire in close cooperation with a programmer, a

human computer interface specialist, and a project manager on site at Icon
Medialab. In essence, the construction of this questionnaire was a mini-
internet consulting project and was a large undertaking due to the creation
and programming of a special section for the social network data
collection. We placed the questionnaire on the company’s intranet as well
as linked it directly to the company’s personnel roster for an up-to-date
listing of all active, full-time employees across the company’s 26 offices in
16 countries Australasia, Europe, Asia, and North America. Thus, when
each individual was responding to the questionnaire, he or she could easily
surf through the various offices and names to click on the appropriate
individuals with whom he or she communicated on advice-related matters.
In this manner, we could collect data on an individual’s participation in the
various networks of the entire multinational.
We placed a hyperlink to the survey at the top of the company’s intranet

homepage so that individuals could easily find the survey. In addition,
there were hyperlinks to the survey within the introductory email as well as
within all reminder emails. Due to the length of the survey, we designed
the survey so that when the individual moved from one section to the next,
answers were automatically saved in the survey database. In this manner,
an individual could leave the survey and return at any time through the
intranet link to find his or her previously entered answers. After pilot
testing the survey with 15 people in 15 offices, we made several changes to
avoid misinterpretations of the questions as well as to remove several
technical bugs in the survey that were generally caused by different
operating environments. For example, we found that the survey worked
very well using Microsoft’s Internet Explorer on PCs but not on Macs.
Because our research required the complete network, we had to specify a

boundary around it. We used the membership criterion (Marsden, 1990;
Wasserman & Faust, 1994:31), thus we included those individuals who
were formally employed and active in the organization. Individuals who
were currently on leave of absence, working only part-time, or were
independent consultants working for the company were eliminated from the
respondent pool since their networks would not be comparable to those
employees who were actively working full-time for the organization. The
resulting number of total potential respondents was 1698.
Since achieving an 80% response rate is difficult in any research situation,

several measures were taken. First, all respondents were entered into a
drawing for 14 prizes of approximately US $1600 in total value. Second,
several mailings were sent out by email to each individual, including 1) an
initial request by email for participation from the CEO of the appropriate
office, 2) a request from the researcher by email, 3) a personal follow-up by
email two weeks after the first mailing, 4) and if necessary, a second and
third personal follow-up by email three weeks after the first mailing.
Third, we tried to be responsive to respondents throughout the course of
data collection. Once a survey was published on the intranet, we actively
monitored our email inbox throughout all waking hours due to the many
different time zones represented by the company. Whenever a question or
concern came in, we tried to immediately respond to the person sending the
mail. In addition, by having our email placed on an individual office’s
mailing list, we were able to follow any discussions taking place
electronically between the unit’s employees. In this manner, we were able
to immediately answer any questions as well as extinguish or at least
dampen any fires that were raised around the survey. As a result, in the
course of the project, we sent and received more than 5000 emails, with
some of the ones received being of a rather scathing nature due to the
sensitive nature of the social network questions. Finally, throughout the
data collection process, individuals were assured that their responses would
be kept confidential on a secure server at the company’s third party intranet
host and that results would only presented in aggregate form.
We received 1439 completed surveys for a response rate of 84.7%,

comfortably above the 80% cutoff level. To the best of our knowledge, a
survey this comprehensive in global reach is the first of its kind – an
intranet-based sociocentric (i.e., network questionnaire administered to all
146 CHAPTER FIVE
employees within one organization across the globe). To have performed

this survey prior to the advent of internet-based communication would have
virtually been impossible. The survey would have been at least 50 pages
thick due to the names of all the employees, quite a daunting document to
ask people to complete, not to mention the resources that would be needed
to spend on both administering the survey around the world and inputting
the endless questionnaire and network data into a database.
To analyze this large amount of data collected, we used UCINET software

(Borgatti et al., 1999) to create social network measures for each of the
individuals. What we thought would be a relatively easy task turned into
one that took several months just to get to the point where we could begin
analyzing the data. Due to the demands of social network analysis and our
research model, we had to ensure that the database only included active,
full-time employees. While the database was based on the company’s
employee roster, we discovered that this list also included the names of
individuals who were on leave of absence, worked part-time, or for some
other reason were not appropriate for the survey. Thus, the first task was to
remove approximately 300 individuals from the database. The next step
was to ensure that we had the correct background data for each of the
remaining 1698 employees, e.g., hierarchical position, functional
competence, etc., and that the information for each of the categories was
consistent across the population. While we had programmed the survey to
automatically retrieve the individual’s title and competence data from the
company’s intranet, we soon learned that these data were not to be
completely trusted. A quick inventory revealed that there were more than
1100 different titles at Icon, and while even though two titles may have
been the same across units, the actual task activities of the individuals
could still differ across units. Thus, our next task was to return to the
organizational data collected from each individual office as well as search
the company’s intranet and research each and every one of the 1698
individuals in order to determine the correct background data for each one.
When these data were finally updated, we created social network measures
for each individual before importing them into the questionnaire database.
We then analyzed the data through structural equation modeling using the
EQS 5.7 software.
5.3.3 Research at Three High Technology Firms – Article 7

Since the purpose of this final study strays from the previous four articles
and is to explore knowledge dissemination in networks of practice in a
multinational setting, we decided to focus on a limited number of MNCs.
We based the selection criteria on a number of factors: 1) annual sales
greater than USD 15 billion, 2) large, globally dispersed R&D operations,
and 3) operating in the high-technology electronics sector. We chose three
companies: Hewlett Packard (HP), one other US-based company (A), and
one Swedish-based company (B). The latter two are disguised, according
to the wishes of the companies.
While explicitly conducted for this thesis, this study diverts from the
previous studies and adopts a case research approach to the empirical
investigation because of the importance of studying knowledge flow
processes in their real-life context (Yin, 1989). This approach is
particularly important given our emphasis on studying what actual
mechanisms are being employed for knowledge dissemination, rather than
the mechanisms intended for knowledge dissemination by top management.
A secondary reason for choosing a case study approach is that we felt the
existing body of literature did not adequately describe the phenomenon
under investigation. As stated by Eisenhardt (1989a:548), “There are times
when little is known about a phenomenon, current perspectives seem
inadequate because they have little empirical substantiation, or they
conflict with each other or common sense…..In these situations, theory
building from case study research is particularly appropriate.”
At each of these three companies, we conducted ten to twenty-five in-depth

field interviews from June 1997 to February 1998 for a total of fifty-five
interviews. People at different levels of the company: corporate R&D
manager, laboratory manager, project manager, and researcher, were
interviewed for one-and-a-half to two hours each. We took several steps to
increase the reliability and validity of the results. For example, two
interviewers were present at all the interviews that were based on a semi-
structured interview guide. Also, each interview was taped and transcribed
by one of the interviewers. Immediately following each interview,
interviewers discussed individual impressions and differences were
resolved. Some written material was also collected from the companies.
The data analysis then proceeded through several stages. First, the
148 CHAPTER FIVE
interview data were reduced and classified before they were analyzed for
commonalties based on knowledge dissemination.
5.4 Summary
As can be seen in table 5.1, the empirical studies involve a wide variety of
data collection, and the methods used are quite varied. While all
individuals involved in these studies can be defined as knowledge workers,
they come from a wide range of occupations, e.g., programmers, lawyers,
construction engineers, office administration, etc., and are located across
the globe in numerous countries, e.g., the U.S., Singapore, France,
Australia, etc. Thus, this approach of using multi-methods and multi-sites
has substantial advantages since it allows for the development of a richer
understanding of networks of practice and facilitates generalization. We
provide an overview of the different research questions, sites, and methods
in table 5.1.
Table 5.1 Overview of Research Sites and Methods

Art. Purpose Research Site Methods
1 RP 1a: To conceptually -Sundlink, international -Paper-based survey,
develop the structural consortium led by n=120 (87.6%)
properties of Skanska AB -Social network analysis
communities of practice -All individuals in four -Correlation analysis
and propose a series of departments
relationships between -Construction engineers
structural properties and
performance
2 RP 1b: To investigate -US professional legal -Bulletin board messages
the structural properties association -MS Word survey as
of an electronic network -All members of email attachment, n=152
of practice through the association’s electronic (29%)
application of theories of network of practice -Text analysis
collective action and -Lawyers -Social network analysis
public goods -Correlation analysis
3 RP 2: To investigate the -Icon Medialab -30 interviews

relationship between -All employees -Paper-based survey,
individual participation -Nine offices in eight n=203 (84%)
in various types of European countries -Correlation analysis
networks of practice and -Variety of professions: -Multivariate regression
individual performance management, analysis
programming, art
direction, etc.
4 RP2 -Cap Gemini -5 interviews
-40 offices in three -Html survey as email
Nordic countries attachment, n=83 (24%)
-All 345 members of the -Correlation analysis
NCN MS electronic -Text analysis of open-
network of practice ended survey questions
-Programmers
5 RP2 -Cap Gemini -5 interviews
-40 offices in three -Html survey as email
Nordic countries attachment, n=83 (24%)
-All 345 members of the -Partial least squares
NCN MS electronic analysis
network of practice
-Programmers
6 RP2 -Icon Medialab -35 interviews
-26 offices in 16 -Internet-based survey,
countries in Asia, n=1439 (84.7%)
Europe, Australasia, -Social network analysis
North America -Correlation analysis
-All employees -Structural equation
-Variety of professions: modeling
management,
programming, art
direction, etc.
7 RP2 -R&D operations of three -55 interviews
high technology -Text analysis
multinationals (two
European and one US)
-Scientific researchers
5.5 Generalizability
In any research study, the important question that must be asked is to what
degree the results from the research can be generalized for a larger group of
people than those who participated in the study. As mentioned, all
respondents in the studies were conducting knowledge work in
150 CHAPTER FIVE
organizations that were knowledge-intensive. In general, it is difficult to

assess how representative the findings are for other organizations and
networks of practice, especially those that are of a less knowledge-intensive
nature. However, it is difficult to generalize the results for the studies that
address Research Purpose 1. The study on the community of practice was
conducted at only one site, the international construction consortium, and
for the study of the electronic network of practice of US lawyers,
examination is restricted to one type of electronic network of practice
within one industry. However, participants in this study are from multiple
organizations, helping to ensure generalizability to networks of practice
that span organizational boundaries.
With regard to the studies at the individual level, the ability to generalize
across knowledge-intensive organizations is supported to some extent since
we find similar trends across the two different organizations studied despite
one being a well-established, more conservative, and large multinational
and the other a young, medium-sized multinational. In addition, the results
from the extensive social network study at Icon are based on responses
from employees in different age groups, of different gender, in different
hierarchical positions, of different educational backgrounds, in different
functional tasks, and from 16 countries across North America, Europe,
Australasia, and Asia. Thus, the results can be seen as tentative indications
of the relationships between participation in different networks of practice,
centrality, and individual performance. Finally, the ability to generalize the
results in the last article that was based on case studies is quite limited due
to the more exploratory nature of the study.
In summary, although the focus was on individuals exchanging knowledge

while conducting knowledge work, networks of practice within
organizations as well as across them may have different membership and
exchange dynamics. Thus, studies comparing various networks and their
network dynamics both within and across firms as well as those publicly
available on the internet are suggested.
CHAPTER SIX
Summaries of the Empirical Studies
THIS CHAPTER PROVIDES a summary for each of the seven

empirically based articles presented in Appendix Two. For the studies and
findings based on qualitative data, we will present a lengthier summary
here. However, for those studies based on an analysis of quantitative data,
we will present a brief summary of the results and leave the discussion for
the next chapter, Discussion and Implications.
Article 1
Theorizing Structural Properties of Communities of Practice: A Social

Network Approach
By A. Schenkel, R. Teigland, & S. P. Borgatti
Previous version presented at Academy of Management, 2001.
Article 2
The Provision of Online Public Goods: Examining Social Structure in

a Network of Practice
By M.M. Wasko & R. Teigland
Nominated runner-up Best Paper ICIS 2002
Previous version published in The Proceedings of the 23rd International

Conference on Information Systems (ICIS), Barcelona, Spain, 2002.
152 CHAPTER SIX
Article 3
Communities of Practice in a High-Growth Internet Consultancy:

Netovation vs. On-Time Performance
By R. Teigland
Versions published in three books:
In E. L. Lesser, M.A. Fontaine, & J.A. Slusher (eds.), Knowledge and

Communities, Boston: Butterworth-Heinemann, 2000.
In J. Birkinshaw & P. Hagström (eds.), The Flexible Firm, London: Oxford

University Press, 2000.
In F. Delmar & P. Davidsson (eds.), Tillväxtföretagen (High-Growth

Firms), Stockholm: SNS Förlag, 2001 (in Swedish).
Article 4
Extending Richness with Reach: Participation and Knowledge

Exchange in Electronic Networks of Practice
By R. Teigland & M. M. Wasko
To be published in P. Hildreth & C. Kimble (eds.), Knowledge Networks:

Innovation Through Communities of Practice, London: Idea Group Inc.,
2004 (forthcoming).
Previous version published in W.J. Orlikowski, S. Ang, P. Weill, H.C.

Krcmar, & J.I. DeGross (eds.), The Proceedings of the 21st International
Conference on Information Systems, Brisbane, Australia, ICIS, 2000.
EMPIRICAL STUDIES 153
Article 5
Integrating Knowledge Through Information Trading: Examining the

Impact of Boundary Spanning Communication on Individual
Performance
Version published in Decision Sciences, Special Issue on Knowledge

Management, 2003 (forthcoming).
Article 6
Exploring the Relationships Between Network of Practice

Participation, Centrality, and Individual Performance in a
Multinational Organization
By R. Teigland
Previous version presented at INSNA – International Network of Social

Network Analysts Sunbelt Conference, 2002.
Article 7
Knowledge Dissemination in Global R&D Operations: An Empirical

Study of Multinationals in the High-Technology Industry
By R. Teigland, C.F. Fey, & J.M. Birkinshaw
Published in Management International Review, Special Issue on

International Management of Technology, 2000, 1.
6.1 Article 1. Theorizing Structural Properties of

Communities of Practice: A Social Network Approach
As revealed in the review of Figure 6.1 Positioning of Article 1

empirical studies, research on
communities of practice is very Face-to-face Mixed Electronic
biased towards ethnographic

Co-located
RP1a:
research investigating the cognitive Article 1 0/ 0/
aspects of communities of practice

Non-co-located
within co-located groups of 0/

individuals. This literature has
concentrated on defining
organizational
Inter-
communities of practice, primarily

focusing on how communities
emerge and operate and often
relying on anecdotal accounts as the
basis for theory development. Additionally, researchers have paid little
attention to conceptualizing the structural properties of communities of
practice or the relationship between communities of practice and
organizational performance. These gaps in the research seem surprising
because on the one hand, there is such a strong relationship between
cognition and structure, and on the other hand, the emergent structure has
been shown to play a strategic role in organizational outcomes. Thus, this
article investigates Research Purpose 1a: to conceptualize the structural
properties of a community of practice and propose a series of relationships
between community of practice structural properties and performance.
Figure 6.1 depicts the positioning of Article 1 within the network of
practice matrix.
In this article, concepts are drawn from the social network literature and
synthesized with the existing literature on communities of practice. Based
upon this synthesis, we propose four structural properties for a community
156 CHAPTER SIX
of practice on the community level (connectedness, graph-theoretic

distance, density, and core/periphery structure) and one structural property
on the individual level (coreness). These structural properties are presented
in table 6.1
1) Connectedness. Community of practice researchers see the primary

requisite for the development of a community as the mutual engagement
among individuals (e.g., Wenger, 1998). As the ethnographic studies have
shown, community of practice members engage in a fluid stream of
collaboration and narration, helping each other to perform their tasks. For
example, Wenger’s claims processors talked and interacted at their desks as
they worked while Orr’s technicians told war stories over breakfast. The
result of this interaction is a complex network of social relations and
interdependency. Thus, we argue that the extent to which individuals are
connected via pair-wise interaction ties is an index of the extent to which
they can potentially function as a community of practice. Individuals who
are not interacting with others in a group cannot learn the community’s
practice and thus will not be identified as being members of the
community. Therefore, a minimum structural characteristic of a
community of practice is that every member has appropriate ties (e.g.,
advice-giving, trust, etc.) with some if not all other members of the
community. In other words, all community members are directly or
indirectly connected with each other and there are no isolates. In social
network analysis, the maximal set of individuals who are directly or
indirectly connected to each other in a network is called a connected
component (Harary, 1969). Therefore, a community of practice is
necessarily located wholly within a single connected component.
2) Graph-theoretic Distance. Another fundamental characteristic of

communities of practice is the notion of shared repertoire of both a tacit
and explicit means of communicating and working that includes language
and unarticulated etiquette (Brown & Duguid, 1991; Wenger, 1998).
Social network research has studied the diffusion of sharing of ideas and
attitudes extensively (Friedkin, 1982; Burt, 1992; Rogers, 1995). A central
tenet of this research is the notion that in both diffusion and influence
processes, the graph-theoretic distance between nodes in a network dictates
the extent to which they are expected to share ideas. The graph-theoretic
distance between two nodes is defined as the number of links in the shortest
path connecting them. Thus, the greater the graph-theoretic distances
between pairs of group members, the longer it takes for information to flow
from one to the other, and the greater the likelihood that what is transmitted
arrives too late, too distorted, or fails to arrive at all. Individuals separated
by wide distances then tend to develop variations, e.g., in language, values,
norms, etc., that are not shared, contrary to the notion of a single
community of practice. Thus, we would expect that the average graph-
theoretic distance between all pairs of members of a community of practice
would be shorter than the average graph-theoretic distance between all
pairs of individuals within organizational networks in general.
3) Density. Through mutual engagement and the associated cognitive

processes, the practice of a community is disseminated and developed.
Connectedness is a necessary prerequisite for this development but not
sufficient in and of itself since a certain level of density is required. The
density of a network measures the degree of cohesion in the group (Blau,
1977) and is defined as the total number of ties divided by the total number
of possible ties in the network. A dense network consists of people who
are for the most part directly connected to each other, rather than connected
through intermediaries. Direct connections are far more powerful in terms
of influence and transmitting tacit knowledge. Hence, through a dense
network, a community's practice is more evenly disseminated. In social
network terms, density is a function of the average number of contacts that
each individual possesses, and it is the average number of ties per person
divided by N-1, where N is the number of individuals in the network. A
community of practice should exhibit a higher density than the
organizational network in which it is embedded, which is discussed further
in Structural Property 4.
4) Core/periphery Structure. Community of practice theory distinguishes

between communities and constellations (Wenger, 1998). A constellation
is a set of different communities of practice (possibly involving
overlapping membership) that have different shared repertoires and
different joint goals. Groups that have largely different memberships,
interacting primarily within-group rather than with members of other
groups, and developing separate sets of shared repertoire, can be seen as
forming a single constellation, but not a single community of practice.
Structurally then, it is obvious that communities of practice do not contain
significant subgroupings since such subgroupings would constitute
separate, although interlinked, communities of practice.
158 CHAPTER SIX
In social network theory, a network has a core/periphery structure to the

extent that it contains no significant subgroups, factions, or cliques except
the core itself (Borgatti & Everett, 1999; Everett & Borgatti, 1999). Stated
in another way, a network has a core/periphery structure if it “can be
partitioned into two sets: a core whose members are densely tied to each
other, and a periphery whose members have more ties to core members
than to each other” (Everett & Borgatti, 1999: 397.) Network researchers
have developed statistical procedures for measuring the extent to which an
observed network conforms to a core/periphery structure (Borgatti &
Everett, 1999).
Core/periphery structures facilitate the diffusion of information and

innovation because they do not contain significant clusters of nodes that are
poorly connected to the rest of the network. Consequently, they can be
expected to lead to a relatively homogeneous group culture (a shared
repertoire) in which most individuals are exposed to new practices and
ideas soon after they emerge. In contrast, networks that are divided into
cliques or factions work against the establishment of a single community of
practice. Different subgroups tend to develop their own norms, beliefs, and
practices, which then effectively create separate communities of practice
that are loosely connected to each other – i.e., constellations. This in turn
implies that communities of practice have a core/periphery structure as is
described in social network theory.
5) Coreness. As described above, Wenger (1998) distinguishes full

participation in a community of practice from legitimate peripheral
participation and marginal participation, and the distinction between the
latter two depends on the legitimacy of the individual. This distinction is
fundamentally cognitive rather than structural. Thus, in a network analysis
of a set of relations at a single moment, it would be difficult to distinguish
between legitimate peripheral participation and marginal participation.
However, the difference between these and full participation can be
detected by the coreness measures that are produced as a by-product of
fitting the core/periphery model (Borgatti & Everett, 1999). Technically,
coreness is defined as the principal eigenvector of the network matrix
(Bonacich, 1972). In non-mathematical terms, coreness indicates the
extent to which a node is located in the center or periphery of a group.
Nodes with high coreness are well connected to both core and peripheral
members while nodes with low coreness are connected mostly to core
members. Thus, this structural property mimics the position of new

apprentices in a community, who initially are connected through a few
experienced members who show them the ropes, and who gradually form
ties with more and more people. Hence, coreness is the basis for our last
structural property.
Table 6.1 Structural Properties of Communities of Practice

Structural
Property Description
1. Connectedness In a community of practice, every member is connected, directly
or indirectly, to every other member. That is, a community of
practice is contained within a connected component.
2. Graph-theoretic Relative to organizational networks in general, communities of

Distance practice have shorter graph-theoretic distances between all pairs
of members.
3. Density Relative to organizational networks in general, communities of
practice have a greater density of ties.
4. Core/periphery Communities of practice have core/periphery structures rather

Structure than clique structures.
5. Coreness The greater an individual's participation in a community of

practice, the greater is his or her coreness score.
To illustrate the structural properties of communities of practice, these

properties were applied to the Sundlink bridge project. As described in
Chapter Five, data were collected through a paper-based questionnaire.
Based on our expectations, we found very weak support for considering the
project as a whole to display the structural properties of a community of
practice. While the project does fulfill the structural property of coreness,
it fulfills the next three structural properties to a very low degree. In
particular, there is little support in terms of density that the project network
was a community of practice. This may be explained by the physical
layout of the project: geographically separated sites and uneven resource
allocation that meant that not all the people in the respective departments
could interact on a face-to-face basis. This finding thus supports our
argument that face-to-face interactions are important for the development
of communities of practice and that communities of practice primarily
develop within co-located groups of individuals.
160 CHAPTER SIX
We then applied the properties to the five individual departments within the
project to determine to what degree these groups of co-located individuals
displayed community of practice structural properties. Of the five
departments within the project, only the Technical Department shows
characteristics of a community of practice (figure 6.2). Despite including
individuals with different operative backgrounds, a community of practice
appears to have developed within this department during the course of the
project. However, none of the other departments displays the structural
properties of a community of practice. These findings suggest then that
while the formal organization may impact the development of communities
of practice, it does not necessarily coincide with them, thus also confirming
previous community of practice research.
Figure 6.2 Technical Department
Note: Node shape indicates section membership.
A further analysis at the individual level reveals that coreness is not related
to age, years of experience in the construction industry, or years of
experience in similar duties elsewhere. However, it is significantly related
to the number of years in the current position, the level of education, and
time spent at the office vs. at the construction site. In addition, within one
department, the Technical Department, we find that individual coreness is
related to hierarchical position since most of the core individuals are
section heads. These findings are all aspects that make sense within the
community of practice literature. We would expect the relationship with
tenure in current position because the longer a person participates in a

community of practice, the more opportunity there is for building
relationships and moving from the periphery (where all newcomers begin)
into the core. In terms of education, this may be partly due to
management’s preference for hiring more educated people for central
positions, requiring a high degree of theoretical technical knowledge, but it
also may be that more educated people were considered to be more
knowledgeable than others and, therefore, more often approached for help.
Finally, the negative relationship with time at the construction site also
makes sense in light of the community of practice literature since physical
proximity and thus face-to-face communication are argued to be important
facilitators of mutual engagement.
In addition, we developed three propositions regarding the relationship

between community of practice structural properties and performance. At
the community level, we propose that knowledge sharing within the
community of practice is contingent upon the density and size of the
community and that the ability of the community to solve problems is
contingent upon the complexity of the problem and the variance in the
coreness among the members. Finally, at the individual level, we propose
that individuals who have higher coreness scores have a higher degree of
individual performance. Individuals with lower coreness scores have less
opportunity to gain knowledge from others in the community, resulting in
less community-specific knowledge and a more idiosyncratic practice. At
the same time, their lack of connection with others makes them less
influential and less able to shape the community's practice.
6.2 Article 2. The Provision of Online Public Goods:

Examining Social Structure in an Electronic Network of
Practice
Nominated runner-up Best Paper at ICIS 2002

Conference on Information Systems (ICIS), Barcelona, Spain, ICIS, 2002.
Electronic networks of practice are Figure 6.2 Positioning of Article 2

similar to communities of practice
in that they are a social space Primary Communication Channels

where individuals working on
Co-located
similar tasks self-organize to help 0/ 0/

each other and share perspectives
about their occupational practice

Non-co-located
or common interest. However, 0/

unlike communities of practice and
the other types of networks of
organizational
RP1b:
Inter-
practice where people know each Article 2
other personally and form dyadic,

interpersonal relationships,
participants in electronic networks
of practice are typically strangers,
individuals form weak ties with the entire network instead of with a select
few, and interactions occur through text-based, often asynchronous,
computer-mediated communication. More importantly, the differences
between an electronic network of practice and the other types of networks
of practice relate to the visibility of the interactions between members. In
electronic networks of practice, such as those supported by listserv and
bulletin board technologies, the ability to post messages is generally open
to anyone and once posted, messages are visible to everyone participating
in the network. Due to this open nature of this type of network of practice,
some researchers have suggested that theories of collective action and
164 CHAPTER SIX
public goods be applied to further our understanding (Steinfield, 1992; Fulk

et al., 1996; Kollock & Smith, 1996).
From the viewpoint of the thesis as a whole, this article contributes to our
understanding of electronic networks of practice by investigating the
structural properties of an electronic network of practice through the
application of theories of collective action and public goods. Addressing
Research Purpose 2, this article not only fills the gap relating to structural
properties of networks of practice, but it also contributes to the body of
empirical studies on electronic networks of practice. As we saw in the
review of empirical studies on electronic networks of practice, there was a
dearth of studies focusing on electronic networks of practice, be they inter-
organizational or intra-organizational. Figure 6.3 depicts the positioning of
Article 2 within the network of practice matrix.
In order to achieve this purpose, this article builds on the work conducted
in Article 1 on structural properties of a community of practice. However,
due to the distinct world of online interactions, we found that we had to
adapt the structural properties created for communities of practice to
electronic networks of practice. The social network measures developed
for the community of practice study were partly based on identifying and
measuring the presence of interaction between two individuals. However,
since postings of messages are visible to everyone in an electronic network
of practice based on listserv or bulletin board technologies, all members
interact automatically with everyone else and are only “one click away”
from each other. While variations may occur in electronic network of
practice formats, knowledge is generally visible to all other members and is
automatically shared with all others. As a result, the measures of
connectedness, graph-theoretic distance, and density that were developed
for communities of practice are not as relevant in an electronic network of
practice setting since individuals are automatically interacting with all other
members due to the nature of the electronic network of practice. A
different approach is then needed in order to investigate the structural
properties of an electronic network of practice.
Thus, heeding the call by previous researchers, we reviewed the literature

on collective action and social dilemmas to facilitate the development of a
series of structural properties specifically for electronic networks of
practice. Through this process, we developed three research questions
relating to the structure and knowledge flows of an electronic network of

practice (provided below). We examined these three questions through the
analysis of an inter-organizational electronic network of practice of a US
professional legal association. The shared practice of this network was US
federal law, where participants (lawyers) actively engaged in exchanging
legal advice in the electronic network of practice. We collected data
through the use of a MS Word questionnaire distributed as an email
attachment as well as the text analysis of 2460 messages posted to the
network during a two-month period. We examined all electronic network
of practice messages to determine the identity of the person posting, and we
then coded each one as a seed (the first message in a thread), a singleton
(seeds without responses), a question, a response, or other. We then built a
social network matrix consisting of all 526 participants to determine who
was responding to whom, creating a directed, social tie.
As described in Research Purpose 1b, this study focuses on the production

of knowledge as a public good in electronic networks of practice. In the
formal language of collective action theory, the network participants are the
interest group and the public good is the continuous stream of knowledge
produced and jointly held by the network’s participants. We argue that the
knowledge produced by the electronic network in this study is a public
good since it exhibits the characteristic of non-rivalry and non-
excludability. A brief summary of the results for each research question is
presented.
RQ1. What is the pattern of contribution that produces and sustains the
network of practice public good?
The first key issue for examination is the pattern of contributions that
creates the public good. In electronic networks of practice, contribution is
reflected in the posting of questions and replies that take the form of a
conversation. This interaction creates social ties between participants. We
define a social tie in an electronic network as the tie created between two
individuals when one person responds to another’s posting. While it has
been argued that social ties are important for collective action, it is less
well established as to exactly how and why social ties are important
(Marwell & Oliver, 1988). Collective action theory provides three views
regarding the pattern of contributions or social ties that is necessary to
create the public good: a dense network consisting of direct ties between all
166 CHAPTER SIX
members (Marwell et al., 1988), dyadic reciprocal exchange (Kollock,

1999), and generalized exchange (Fulk et al., 1996).
Our results reveal that people do not post an equal number of messages to
the electronic network of practice, indicating that members do not
participate equally in the provision of the public good. In addition, the
knowledge flows in this network of practice are characterized not by dyadic
exchange but by patterns of generalized exchange, i.e., one’s providing of
help is not reciprocated by the recipient, but by a third party. Thus, our
results are in line with the work by Fulk et al. (1996).
RQ2: Are networks of practice characterized by a critical mass

constituting a core?
Borrowing from nuclear physics, the theory of critical mass argues that a
subset of a group may be responsible for making the majority of the
contributions to the production and maintenance of the public good (Oliver,
Marwell, & Teixeira, 1985). As in Research Question 1, this property is
examined by looking at the pattern of social ties. The presence of critical
mass is determined by the degree to which ties are centralized or
concentrated to a few individuals rather than spread across the entire group.
As mentioned above, building upon Wenger’s categories of community of
practice participation and based upon the analysis of messages, we created
four categories of participants: outsiders (people who posted seeds, but
never received a response), seekers (people who posted only questions),
periphery (people who posted 10 or less responses) and insiders (people
who posted more than 10 responses). Using UCINET software (Borgatti,
Everett, & Freeman, 1999), we analyzed the ego network of each
individual to determine centrality in terms of “in degree”, i.e., the number
of times other people respond to an individual, and “out degree”, i.e., the
number of times an individual responds to others. Through this analysis,
we found that the network is structured as a star with a critical mass
surrounded by peripheral connections emanating outwards. We also
performed a component analysis, which revealed that the electronic
network of practice is characterized by only one component and not a set of
subsets. This indicates that there are no cliques, rather the critical mass
actively responds to many unique and overlapping individuals, and the
periphery engages in both receiving and providing advice to others.
However, contrary to community of practice theory, individuals forming
the critical mass are not tied to one another. Using Krackplot, figure 6.4
shows the network structure of survey respondents (Krackhardt et al.,
1994).
Figure 6.4 Structure of Electronic Network of Practice
RQ3: How does the heterogeneity of resources and interests of

participants impact network of practice collective action?
A population’s heterogeneity of resources and interests is argued to affect

collective action and the production of a public good (Olson, 1965; Hardin,
1982; Oliver et al., 1985). The more heterogeneous a group is, the more
likely there is a critical mass or subset of members who have a high enough
level of resources and/or interests to produce the public good. However,
heterogeneity can also hinder collective action even when the mean levels
of heterogeneity appear sufficient. As such, the distribution of
heterogeneity is important in terms of collective action, i.e., the more
positive skew and deviation from the mean, the more likely a critical mass
may result (Oliver et al., 1985). Resources include money, time, expertise,
energy, and influence (ibid). We examined the importance of resources
168 CHAPTER SIX
and interests by analyzing the correlations between network centrality data

and survey measures of resources and interests. Results indicate that
interests and resources are not as significant for people who receive help,
but they are reasonably good indicators of why people provide knowledge
to others. The only significant relationships with receiving help (in degree)
are sustainability and challenge, thus those who received help were
interested in continuing their electronic network of practice participation
and the challenge associated with doing so. The results indicate that longer
professional association tenure and higher levels of expertise are associated
with responding to others. In addition, individuals who were sole
practitioners were significantly related to responding to others as were
those concerned with enhancing their reputations.
6.3 Article 3. Communities of Practice in a High-Growth

Internet Consultancy: Netovation vs. On-Time Performance
By R. Teigland
Versions published:


In F. Delmar & P. Davidsson (eds.), Tillväxtföretagen (High-Growth

Firms), Stockholm: SNS Förlag, 2001 (in Swedish).
This article is the first of five that Figure 6.5 Positioning of Article 3
address Research Purpose 2, and it
was also the first one conducted in Face-to-face Mixed Electronic
this vein, thus it is of a more

Intra-orgnizational
Co-located
RP2:
exploratory and empirical nature. Article 3 0/ 0/
The findings from the study of the
Non-co-located
patterns of individual-level RP2:

knowledge flows at Icon Medialab
0/ Article 3
and the impact of those patterns on

organizational
individual performance are RP2: RP2:

0/
Inter-
Article 3 Article 3
described. Building on the
knowledge-based view of the firm
literature, and specifically the
work concerned with networks of
practice, we developed a series of propositions linking individual
performance to various sources of knowledge. This article is positioned in
the network of practice matrix as illustrated in figure 6.5. Research was
conducted through interviews and a paper-based questionnaire (203
responses, 84% response rate, in nine offices in eight countries).
170 CHAPTER SIX
The primary intent of this exploratory study is to understand what

knowledge sources individuals use in their everyday work in an internet-
intensive environment and what role the internet and networks of practice
play. The secondary intent is to take this research one step further by
linking an individual’s knowledge seeking behavior to an individual’s
work-related performance. In contrast to previous research on individual
performance, we break down performance into two components – on-time
(efficient) and creativity. On-time performance is defined as the ability to
achieve targets and objectives as defined by one’s boss or superior. While
there are many views of what creativity is, e.g., “the process of using
imagination and skill to invent a unique product or thought” (Scott,
1995:66), the definition used in this thesis is the ability to develop
innovative solutions to work-related problems in the course of one’s work.
Interesting patterns of knowledge seeking activities were revealed through

an analysis of the data using stepwise regression analysis and a t-test to
compare the means of knowledge source usage. The more technically
oriented people, e.g., software programmers and system architects, tended
to use external codified sources of information such as internet webpages
and electronic networks of practice to a higher degree than non-technically
oriented people did when solving work-related problems. Not only were
these sources the most frequently used, they were also found to be the most
helpful by the technically oriented people. However, non-technically
oriented respondents interacted with external people such as customers and
friends and with internal distributed networks of practice to a higher degree
than technically oriented respondents. Thus, this suggests that the type of
knowledge source selected (and thus which network of practice) is
dependent on the type of task performed.
Turning to individual performance, taking creativity first, social contact

with internal network of practice members outside of work and the use of
external codified sources of information (electronic networks of practice
and the like) are the significant predictors. These findings are particularly
strong when we take a closer look at the technically oriented group.
Looking at on-time performance, the use of internal codified sources of
information is found to be a positive predictor of on-time performance for
the whole sample. However, the use of external codified sources is a
negative predictor of on-time performance for the technically oriented
group. A summary of the findings is presented in table 6.2.
Table 6.2 Results from Stepwise Regression Analysis

Creativity On-time
Proposition Whole Tech. Whole Tech.
Sample Sample Sample Sample
1. Interaction with internal sources
(community of practice)
2. Social contact outside work with = =
.17 .30
3. Interaction with external sources
(customer, inter-organizational distributed
network of practice)
4. Use of internal codified sources (e.g.,
.23* .55**
intranet)
5. Use of external codified sources
.15= .35* -.31*
including electronic networks of practice
=
p < .10
* p < .05
** p < .01
*** p < .001
In addition to the above findings, the qualitative data reveal some

interesting findings. Based on our interviews, the impression one gets is
that technically oriented employees attach great importance to their
contacts in their external networks of practice as sources of ideas and as
ways of solving tricky problems. Several programmers even stated that
they preferred to go first to their internet community or use their external
private email list for help instead of asking someone at Icon even if he or
she were sitting at the next desk. Several explanations were offered. The
first is that by posting a question in an open inter-organizational electronic
network of practice, people are not obligated to help. Instead those who
want to help can do so in a voluntary fashion. By reaching out to the
electronic community for help, one does not disturb a colleague at work
who has his or her own schedule and deadlines to meet. Another reason is
that people can access a much broader source of expertise than at their own
company. In many instances, individuals claimed that there was no
“critical mass” internally, especially when discussing the intranet, within
Icon. However, this critical mass could be found in inter-organizational
electronic communities. Members of inter-organizational electronic
communities work at different types of companies all over the world, yet
they work on the same type of problem. Thus, interviewees were of the
opinion that participation in an electronic network of practice enabled them
172 CHAPTER SIX
to gain access to the latest thinking within their field, especially since the
change of pace within the internet consulting industry is so rapid.
To turn the discussion to the second independent variable, social contact

outside of work, we found in our interviews that individuals became
members of tightly knit communities of practice through extensive social
contact outside of work. During this social contact, these individuals
discuss the difficult problems encountered during the day, the responses
received from the electronic community, and how they then attempted to
solve their problems. The latest solutions or tips from both the outside
communities and one’s own work are passed between the members of the
community. In this manner, these community members socially construct
their world through the narration of stories, turning incoherent data into
coherent knowledge. This enables them to gain insights into the work they
are performing, allowing them to be more creative in their daily work.
What is interesting here then is the combination of the interaction with an
individual’s external network of practice with an individual’s internal
network of practice. As ideas cross community boundaries, resulting in the
cross-fertilization of communities, knowledge is combined and placed
within the company’s local context to foster creativity.
In terms of achieving on-time performance, a very different picture

emerges. Here, the use of internal codified sources of information is a
positive predictor of on-time performance, while the use of external
codified sources is a negative predictor. This is entirely in keeping with
intuitive expectations. Building relationships with external communities
and creating unique or “elegant” solutions on the basis of those
relationships works well when creativity is the objective, but it is a strong
negative when on-time delivery matters. Gathering knowledge from the
outside takes time because either the sources must be located or one must
wait for someone to volunteer help. And once the knowledge or help is
received, it must be assimilated into the context of both the problem and
the company’s way of doing things. As interviewees indicated, this may
take considerable time depending on the complexity of the knowledge and
the problem.
In addition, we found that reciprocity within these inter-organizational

electronic communities is necessary in order to become a true member. In
other words, to be able to ask the other internet community members for
help, one must prove that one also gives back to the network through
providing help to others when asked. This returning of help then results in
the individual performing work for others outside the company. This then
takes away time from the individual’s internal responsibilities, potentially
leading then to poor on-time performance.
We also investigated the high use of internet webpages and electronic

communities and found that prestige is a significant factor in their use.
Several interviewees commented that some individuals feared making
mistakes or making themselves look stupid by asking others at Icon for
help. So, they turned to the internet where “no one knows if you’re a
monkey”. Another aspect is that interviewees viewed membership in
closed or invitation-only internet communities as being prestigious. In
addition, interviewees indicated that some programmers were under a form
of social pressure from their external network to help fellow members solve
their difficult problems, often attempting to “show off” in front of the
others. This was found to lead to conflicting goals or loyalty for the
programmers. Creating a “cool” solution or trying to impress a global
community through solving another external member’s difficult problem
leads to longer hours worked, using unnecessary resources as well as
causing delays in product delivery to the customer. However, it is this
participation in external networks of practice that leads to the cross-
fertilization of networks that then fosters creativity, an important factor in
the creation of a firm’s competitive advantage.
6.4 Article 4. Extending Richness with Reach: Participation

and Knowledge Exchange in Electronic Networks of Practice
By. R. Teigland & M. M. Wasko

Innovation Through Communities of Practice, London: Idea Group Inc.,
2004 (forthcoming).
This article continues along Figure 6.6 Positioning of Article 4

the same lines as Article 3.
However, it digs more deeply Primary Communication Channels
into investigating an intra- Face-to-face Mixed Electronic

Co-located
organizational electronic RP2:

network of practice since we
Article 4 0/ 0/
know much less about these

Non-co-located
networks of practice RP2: RP2:

compared to traditional, face- 0/ Article
Article 44 Article 4
to-face communities of
practice within organizations.
organizational
Inter-
Thus, the goal of this article is 0/

to examine knowledge
exchange in an intra-
organizational electronic
network of practice as well as the relationship that this participation has
with individual performance. In this study, individual performance is
measured with a focus on creativity. The research site is Cap Gemini
Nordic, and we collected data from software programmers through the use
of interviews and an html questionnaire linked to an SQL database sent as
an email attachment. In addition to survey measures, participants also
provided insights by responding to open-ended questions about their
participation in the electronic network of practice. Figure 6.6 provides the
positioning of this article on the network of practice matrix.
In particular, we examine the relationships between knowledge acquisition

and knowledge contribution, electronic network of practice tenure and
participation, individual performance, and reliance on co-located coworkers
176 CHAPTER SIX
(indicating community of practice participation). Our results indicate that

higher levels of participation and tenure in the intra-organizational
electronic network of practice are associated with both acquiring
knowledge from participation in the network of practice and contributing
knowledge to other network members. In addition, both knowledge
acquisition from and knowledge contribution to the electronic network of
practice are positively related to individual performance. However, tenure
in the electronic network of practice is not associated with higher rates of
participation or with individual performance. Finally, the results show that
individuals who relied on their co-located colleagues for help or advice
with their work tasks reported no associations with participation,
knowledge acquisition, or knowledge contribution. In fact, the survey
results indicate that reliance on co-located colleagues is associated with
lower levels of self-reported individual performance. These results are
summarized in table 6.3.
Table 6.3 Quantitative Results from Survey

Scale Std. Cronbach’s
1 2 3 4 5
Range Mean Dev. Alpha
1. Intra-organizational
Electronic Network of
Practice Participation
Level 1-7 2.3 0.82 n/a
Electronic Network of
Participation Tenure 1-50 10.81 11.64 n/a .06
3. Knowledge
Acquisition 1-7 3.62 1.75 0.95 .52** .29*
4. Knowledge
Contribution 1-7 2.34 1.57 0.85 .59** .23* .52**
5. Co-located Coworkers
(Indicating Community
of Practice Participation) 0-3.71 2.99 1.05 n/a .04 .06 .02 .08
6. Individual Performance 1-7 4.30 1.29 0.82 .41** .13 .24* .57** -.24*
* Significant at the p < .05 level, two-tailed
* * Significant at the p < .01 level, two-tailed
In addition, an analysis of the open-ended survey questions suggests that

electronic networks of practice are a valuable complement to communities
of practice. Individuals indicated that the electronic community was an
excellent means of improving their own level of technical competence and
that they learned through their participation by receiving help and

information related to their work tasks. Additionally, they felt that
participation enabled them to keep current with technical developments as
well as to know who was actively working in different areas. In response
to the question as to why individuals help others on the community, the
answers provided included a norm of reciprocity and corporate
responsibility. We also asked whether participation had helped improve
their work performance. Of the respondents, 62% replied that the
community had helped them. We find two categories of answers. First,
participation greatly improves the speed with which participants are able to
solve their problems. Second, individuals are able to learn and receive new
insights from the community. Thus, learning in a network of practice
appears to be on a broader, more general level (learning about new areas
and topics within the practice) while previous research provides evidence
that learning in a community of practice may be of a deeper, more specific
type (learning how to apply the new area to one’s specific task).
6.5 Article 5. Integrating Knowledge through Information

Trading: Examining the Impact of Boundary Spanning
Communication on Individual Performance

This article takes the research Figure 6.7 Positioning of Article 5

conducted in the previous two
articles one step further by Primary Communication Channels
grounding it in the knowledge- Face-to-face Mixed Electronic

Co-located
based view of the firm, and in RP2:

particular Grant’s theory of
Article 5 0/ 0/
knowledge integration. The

Non-co-located
goal of this article is to examine RP2: RP2:

whether individual performance
0/ Article 5 Article 5
in terms of efficient or creative

performance varies as a result
organizational
RP2: RP2:
Inter-
of participation and knowledge 0/ Article 5 Article 5

exchange in internal and
external networks of practice.
Specifically, we developed a
series of hypotheses relating to
the efficiency of integration, the internal flexibility of integration, and the
external flexibility of integration that predict the relationships between an
individual’s participation in various networks of practice, internal and
external knowledge exchange, and individual performance. Thus, we
examine whether individual performance is related to participation and
knowledge exchange in 1) communities of practice, 2) intra-organizational
distributed networks of practice, 3) intra-organizational electronic networks
of practice, 4) inter-organizational distributed networks of practice, and 5)
inter-organizational electronic networks of practice. As such, this article
addresses Research Purpose 2, and it is positioned in the network of
practice matrix as depicted in figure 6.7. The research site is the intra-
organizational electronic network of practice of 345 programmers within
180 CHAPTER SIX
the Nordic operations of Cap Gemini. As mentioned, we collected data

through the use of interviews and an html questionnaire linked to an SQL
database sent as an email attachment.
Results from the PLS analysis are provided in table 6.4. There is evidence
that a high reliance by individual on communities of practice as sources of
help results in lower levels of creativity. This suggests that the knowledge
of an individual’s local community of practice may be largely redundant
and that the use of this local knowledge, although efficient due to a shared
practice, does not appear to positively impact individual performance. On
the other hand, participation in intra-organizational distributed networks of
practice enhances creativity as evidenced by the positive relationship
between internal knowledge trading and both efficient performance and
creativity. Individuals participating in internal distributed networks of
practice are able to act as bridges between local communities of practice,
accessing non-redundant knowledge from other locations and integrating it
with knowledge of their own. Contrary to expectations, we found that
external knowledge trading has no direct relationship to individual
performance. Rather it affects creativity and general performance
indirectly through its influence on internal knowledge trading.
Table 6.4 Results of PLS Analysis

Internal External
Efficient
Knowledge Knowledge Creativity
Perform.58
Trading56 Trading57
1. Co-located Coworkers
(Indicating Communities of -.20* -.15
Practice Participation)
2. Non-co-located Coworkers
(Indicating Intra-
organizational Distributed
.24**
Network of Practice
Participation)
Electronic Networks of .26**
4. Internal Knowledge
.54** .28*
Trading
5. Contacts at Other Firms
(Indicating Inter-
organizational Distributed
.45**
Network of Practice
Participation)
6. Inter-organizational
Electronic Networks of -.06
7. External Knowledge
.45** .09 .17
Trading
R2 .44 .20 .38 .17
* p < .05, two-tailed test
** p < .01, two-tailed test
56
Label has been changed from Internal Information Trading to Internal Knowledge
Trading to reflect the choice of terminology in this thesis.
57
Label has been changed from External Information Trading to External Knowledge
Trading to reflect the choice of terminology in this thesis.
58
Label has been changed from General Performance to Efficient Performance to reflect
the choice of terminology in this thesis.
6.6 Article 6. Exploring the Relationships Between Network

of Practice Participation, Centrality, and Individual
Performance in a Multinational Organization
By R. Teigland
Previous version presented at INSNA – International Network of Social

Network Analysts Sunbelt Conference, 2002.
This article further extends the Figure 6.8 Positioning of Article 6

research on the relationships
between participation in various Primary Communication Channels
networks of practice and individual Face-to-face Mixed Electronic

Co-located
performance by introducing the RP2:

Article 6 0/ 0/
social network measure of
intra-organizational
centrality. We tested the model

Non-co-located
RP2: RP2:
developed in this article using 0/ Article 6 Article 6
questionnaire and social network

data collected from Icon Medialab
organizational
RP2: RP2:
Inter-
after the company had grown to 0/ Article 6 Article 6
1698 employees spread across 28

offices in Europe, the US, and
Australasia, and Asia. As such, it
addresses Research Purpose 2, and it is positioned in the network of
practice matrix as depicted in figure 6.8.
This article combines Grant’s theory of knowledge integration, Hansen’s

(1996) work on knowledge integration in multinationals, and networks of
practice and social network theory. Similar to the previous article, we
developed a series of hypotheses relating to the efficiency of integration,
the internal flexibility of integration, the external flexibility of integration,
and human capital that predict the relationships between an individual’s use
of various types of knowledge sources (networks of practice and codified
knowledge sources), internal and external knowledge exchange, firm
network of practice centrality, and individual performance (creativity and
efficient performance). In addition, we hypothesize that these relationships
184 CHAPTER SIX
differ for three different task groups depending upon the tasks performed:
Commercial and Support Group, System and Software Group, and Design
Group.
Based on the matrix of social network data of all 1698 individuals, we

calculated a measure of overall intra-organizational network of practice
centrality for each individual in the firm, i.e., the degree to which an
individual is central in the network of the firm’s entire set of intra-
organizational networks of practice (be they communities of practice or
distributed networks of practice). We then analyzed the data including this
centrality measure as well as other measures using structural equation
modeling. We discuss the results for the entire sample and each of the
three task groups below; however, a lengthier discussion of the results for
the whole sample as well as the task groups is presented in Article 6 in
Appendix Two.
Our results provide insight into the literature on networks of practice and
knowledge integration at the individual level of the firm. While we did not
find any support for the hypotheses relating to the use of internal codified
sources, we did find strong or moderate support for several of the
hypotheses involving the participation in internal and external networks of
practice, as well as the two human capital variables of education and
experience. What is also interesting is the fact that we see such different
results for the two different dependent performance variables as well as for
the relationships for each of the three task groups.
Efficiency of Integration. Looking at internal codified sources, contrary to

expectations, we did not find any significant relationship between the use
of these and either creative or efficient performance for the sample as a
whole. As for participation in networks of practice, we found the level of
interaction with co-located coworkers as a predictor of efficient
performance. This is in keeping with our a priori expectations. A high
degree of personal interaction with other members of one’s communities of
practice who share the same language should be a highly efficient source of
knowledge. However, we did not expect to find a positive relationship
between interaction with co-located coworkers and creative performance.
Based on theories of weak ties, we had predicted a negative relationship
since it was expected that the knowledge of an individual’s community of
practice is largely redundant, thus hampering the creation of new ideas.
Internal Flexibility of Integration. We found support for the relationship

between overall intra-organizational network of practice centrality and
creative performance. Thus, our results suggest that efficient performance
is dependent upon the ability of an individual to reuse existing local
knowledge without having to tap into the firm’s global knowledge base.
However, the relationship between overall network of practice centrality
and creative performance supports theory that individuals participating to a
high degree in intra-organizational distributed networks of practice are
acting as brokers, indicating that flexible knowledge integration is the
result of the recombination of knowledge found throughout the firm’s
geographically dispersed communities of practice.
In terms of the drivers of overall intra-organizational network of practice

centrality, we did find the expected positive relationships between internal
knowledge exchange and centrality. Mutual engagement involving the
exchange of knowledge with one’s community of practice leads to overall
network of practice centrality in the firm. This finding is in line with
community of practice theory that argues that movement into a central
position within a community is dependent upon mutual engagement
between the individual and others in the community and that this
movement is jointly determined. However, contrary to expectations, we
also find a direct relationship between participation in intra-organizational
distributed networks of practice and centrality. This finding is unexpected
since we argued that internal knowledge exchange should fully mediate this
relationship. Our findings thus indicate that a central position in the firm’s
networks of practice is only partially dependent upon the individual’s
participation in exchange relationships of sharing and receiving knowledge
with others both locally as well as in other locations. Thus, in this study,
interactions with distributed network of practice members are not
dependent upon knowledge exchange relationships to the same degree as
interactions with community of practice members.
External Flexibility of Integration. Turning to external integrative

flexibility and the use of external knowledge sources, our results support
our hypotheses. The use of external knowledge sources does not have a
direct impact on an individual’s performance. Rather it affects creative
performance indirectly through its influence on internal knowledge
exchange and network of practice centrality. In addition to the
hypothesized relationships, we also found a direct negative relationship
186 CHAPTER SIX
between the use of external knowledge sources and network of practice

centrality. Thus, those individuals who merely used external knowledge
sources to a high degree but did not integrate this knowledge with that of
others through internal knowledge exchange remained on the outskirts of
the firm and by implication achieved lower levels of creative performance.
Differences among Task Groups. Finally, we found considerable

differences in the relationships among the variables across the three task
groups. As for the System and Software Group, this group is in strong
contrast to the other two groups of employees in this organization. This
group shows a very high degree of dependency on the use of external
knowledge sources and participation in inter-organizational networks of
practice. While for the entire sample as a whole, a negative relationship
between the use of external knowledge sources and network of practice
centrality is found, this relationship is positive for the System and Software
Group. In addition, the use of internal codified sources leads to a lower
degree of creative performance. One explanation for this finding may be
linked to the nature of the system and software development field. The
pace of change may be so fast within this field that in order for individuals
to keep pace with development, they must rely to a high degree on the use
of external knowledge sources. Knowledge within the firm may quickly
become irrelevant and the reuse of internal knowledge may hamper
individuals’ ability to develop and implement new solutions and processes.
These findings are in line with the argument that system and software
engineering and design is similar to non-software R&D such as that
researched by Allen and colleagues (Hauptman, 1986). One explanation
may be that software engineers working with the same programming
language may easily communicate and share knowledge with others in
inter-organizational networks of practice due to the more universal nature
of the programming language. As a result, external knowledge used by the
System and Software Group does not need to be absorbed through the
combination with internal knowledge accessed in knowledge exchange to
the same degree as the external knowledge used by the Commercial and
Support Group and Design Group.
These findings then provide evidence that the dynamics of knowledge

sharing among network of practice members within and across firms may
be dependent upon the nature of the underlying practice knowledge. More
detail on the individual task group patterns can be found in Article 6 in

Appendix Two.
6.7 Article 7. Knowledge Dissemination in Global R&D

Operations: An Empirical Study of Multinationals in the High
Technology Industry
By R. Teigland, C. Fey, & J. Birkinshaw
Published in Management International Review, Special Issue on

International Management of Technology, 2000, 1.
While this article is the last of the Figure 6.9 Positioning of Article 7
seven empirical studies of this
thesis, it is actually the first study
that was conducted. This research Intra-organizational
Co-located
laid the groundwork for networks of 0/ 0/

practice as the central theme in this
Non-co-located
thesis. The focus of this introductory RP2:

0/ Article 7
research is on an applied question:
How are MNCs managing the
organizational
knowledge flows in their global

Inter-
R&D operations? While this

research did not set out to explicitly
investigate networks of practice,
results revealed that intra-
organizational networks of practice were one of the primary mechanisms
that facilitated knowledge sharing in the companies investigated. In
addition, we found some interesting results regarding performance at the
firm level. Thus, while this article does not explicitly address Research
Purpose 2 on the relationship between participation in networks of practice
and individual performance, it does serve to fill the second research gap of
performance identified in the review of the empirical studies. Thus, we
report this article under Research Purpose 2, and figure 6.9 illustrates how
this article is positioned in the network of practice matrix.
The basic challenge to the management of multinational corporations

(MNCs) is one of maintaining the responsiveness of individual units to the
opportunities and demands of their local environment while at the same
time capturing the latent benefits that a large, global network can confer.
190 CHAPTER SIX
To create a sustainable competitive advantage, knowledge, it is argued,

must be created at a quick pace while it is simultaneously transferred and
applied throughout the MNC’s global operations (Bartlett & Ghoshal,
1989; Hedlund & Nonaka, 1993; Doz &Hamel, 1997). Coordination across
units prevents the duplication of effort while at the same time ensuring the
fastest time to market with a product that customers want. However,
globally dispersed networks of R&D units create significant managerial
challenges to MNCs. The task of efficiently making use of R&D
knowledge becomes more difficult as many MNCs continue to expand their
global R&D operations, and thereby increase the number of geographically
dispersed locations, employees, functions, and external partners. Both the
complexity of the network and the differences in language and culture lead
to significant challenges. Thus, as mentioned above, the focus of this study
is on an applied question: How are MNCs managing the knowledge flows
in their global R&D operations? The question is descriptive, but it builds
on our a priori expectation that most firms would like to see an increase in
both the volume and quality of knowledge flows between and within R&D
units. To address the research question, we conducted case studies within
three high technology MNCs.
One key finding in this research is that although management spent

considerable time and financial resources on implementing information-
technology mechanisms for individual researchers to use in their everyday
knowledge sourcing activities, the researchers used these to a very limited
extent. Rather they preferred to use their intra-organizational distributed
networks of practice to search for knowledge for help with work tasks.
Several means of facilitating the development of networks of practice

across the companies were also found. For example, one company had
implemented an advanced form of groupware that was a virtual workspace,
enabling researchers in a network of practice to easily share information,
collaborate on documents, and connect with fellow researchers throughout
the organization through such means as bulletin boards and calendars. Of
interest is that there was no central authority or administration for the
system since it was run and maintained by the network. Individual
researchers determined their own level of participation and could designate
what level of security they desired for the information that they entered,
e.g., ”write-only” or ”read-only” access. In addition, there were some
grass-roots efforts such as technology conferences or technology interest

groups.
We also found that the level of encouragement from management to

participate in intra-organizational distributed networks of practice differed
across the companies investigated from no support to full support with
resources. Of the three companies, we found that Hewlett-Packard had the
most active intra-organizational networks of practice. This seems to be due
to management’s support for these networks through the providing of
resources and by allowing them to be of a grass roots nature. People who
were interested in a technology or a subject were free to participate in
meetings of distributed individuals working on the same kind of task across
the organization. Other means focused on increasing the level of
participation in intra-organizational network of practice included personnel
rotation and cross-laboratory projects. Finally, Hewlett-Packard appeared
to have the highest level of shared identity, language, and values of the
three companies due to management’s extensive efforts.
Regarding Research Purpose 2 and the relationship between intra-

organizational networks of practice and firm performance, we found
support for this relationship. We assessed the three companies on six
factors related to R&D performance, e.g., number of articles published,
time to market, and impact of R&D on emergence of successful products.
Hewlett-Packard had the most balanced performance assessment and on no
measure did they score the lowest. This finding follows logically from our
assessment that Hewlett-Packard had the highest level of internal
knowledge sharing that was supported by the company’s considerable
efforts to encourage participation in intra-organizational networks of
practice and to create a shared identity, language, and values across the
organization. Company A, which had the lowest degree of performance,
exhibited the lowest degree of intra-organizational network of practice
participation and shared identity across the organization. Employees
appeared to identify more with their local unit than with the company as a
whole and this led to the feeling of fiefdoms within the company. There
was a high level of “not-invented-here” as evidenced by the resistance to
ideas coming from outside the division due to the strong feelings of local
identity. This was exemplified by the finding that each unit had developed
its own knowledge sharing tools and was resistant to adopting tools from
other units. In addition, individuals at Company A were less interested in
192 CHAPTER SIX
spending time helping others from other units, and they exhibited a higher
degree of “knowledge equals power”.
CHAPTER SEVEN
Discussion and Implications
THIS CHAPTER DISCUSSES and interprets the results from the seven
empirical studies conducted and is divided into five sections. First, the
main findings from the empirical studies are synthesized, and the
theoretical implications of the major findings are then discussed in terms of
the network of practice and knowledge-based view of the firm literatures.
Third, the practical implications of the major findings are highlighted and
explored. A section on limitations and suggestions for how the results from
this study can be expanded in future settings is presented. The chapter then
ends with the dissertation study conclusion. As mentioned previously, the
empirical studies in their article format are attached in Appendix Two.
7.1 Synthesis of Main Findings
The overarching goal of this dissertation is to improve our understanding of

networks of practice from a business firm’s perspective and in particular to
investigate structure and performance in networks of practice within and
between firms. As such, we will discuss the findings for each of two
research purposes in turn. It is important to note that the findings of the
individual studies are not directly comparable due to differences in
methodology; however, we feel that we can make an additional
contribution to the literature by attempting to synthesize the results.
7.1.1 Research Purpose 1: Describing Structural Dimensions

The first research purpose of this thesis is to describe the structural
properties of networks of practice through the application of social network
analysis. The principal idea behind this research purpose is that the
development of a set of structural properties would facilitate the analysis
and further theorizing of networks of practice. Clearly, any set of structural
194 CHAPTER SEVEN
properties is not sufficient in itself as an indicator of the presence of a

network of practice; however, we argue that certain structural properties do
need to be fulfilled in order for a specific network of practice to develop.
In order to investigate this question, two polar forms of networks of

practice were analyzed: an intra-organizational community of practice and
an inter-organizational electronic network of practice. When conducting
exploratory research, it is suggested that cases representing opposite
situations should be selected. Eisenhardt (1989a: 537) states, “…it makes
sense to choose cases such as extreme situations and polar types”.
Eisenhardt’s statement is relevant to this research since the identification of
similarities between the two networks facilitates the ability to generalize
beyond these two cases, despite the two studies not being directly
comparable. Below is a discussion focused on comparing these two
studies.
In order to address Research Purpose 1a, we reviewed the social network

literature for appropriate measures that could be applied to communities of
practice. Five measures focusing on the presence and frequency of
interaction were selected that then created the basis for five structural
properties for communities of practice: connectedness, graph-theoretic
distance, density, core/periphery, and individual coreness. The primary
idea behind the selection of these measures is that we argue that they are
the necessary, but not sufficient, conditions that need to be fulfilled in order
for a community of practice to be present. The underlying assumption in
the community of practice literature is that through dyadic, frequent
interaction individuals develop a shared identity, language, norms, and
values revolving around a joint enterprise. Thus, the more the network is
characterized by a dense network of individuals directly tied to one another
with a central core of individuals surrounded by peripheral members, the
more likely the network is a community of practice and thus will have a
high degree of sharing and incremental innovation of the community’s
knowledge.
We then applied these measures to a large construction project in Article 1.

Results are compatible with the argument that the formal organization does
affect the structure of communities of practice within the organization;
however, the two are not completely aligned. In this organization,
individuals were geographically separated as defined by the organization’s
DISCUSSION 195
departments, and the organization constrained the availability of resources

for individuals to meet face-to-face between departments. Results indicate
that individuals tended to form relationships only with others who were
generally within the same department; however, each department did not
form a community of practice, despite the small number of members. In
addition, individuals who were in the core of the organization’s network of
practice were those who were physically present in their respective
departments to a higher degree. These findings are compatible with the
previous cognitive arguments (Orr, 1996; Wenger, 1998) that communities
of practice are based on face-to-face interactions and as such generally
consist of individuals who are co-located and not distributed. In addition,
our results are in line with theories of legitimate peripheral participation
(Lave & Wenger, 1991) since individuals who had the highest coreness
scores had a higher degree of professional and network expertise. Finally,
we developed a series of propositions relating various community of
practice structural properties to performance as an area for further research.
When we turned our attention to Research Purpose 1b and electronic

networks of practice in Article 2, we found that the properties developed
for communities of practice in Article 1 could be directly applied to
electronic networks of practice. As mentioned, the social network
measures developed for the community of practice study are partly based
on identifying and measuring the presence of interaction between two
individuals. However, in electronic networks of practice that are based on
bulletin board or listserv technologies, postings of messages are
automatically visible to everyone in the electronic network. As such, all
members interact de facto with everyone else. Thus, the measures of
connectedness, graph-theoretic distance, and density that we developed for
communities of practice could not be applied directly to these forms of
electronic networks of practice, implying that a different approach is
necessary to investigate the structural properties of an electronic network of
practice.
As the literature review revealed, previous research on online

communications suggests that theories of collective action and public
goods could be an appropriate lens for our research purpose (Fulk et al.,
1996; Monge et al., 1998). Thus, we developed a set of three research
questions investigating the application of these theories to electronic
networks of practice and their structural properties. We examined an inter-
196 CHAPTER SEVEN
organizational electronic network of practice (listserv) of US lawyers.

Through the investigation of these research questions, we found support for
the application of collective action and public goods theories to electronic
networks of practice. First, we found that the members of the collective,
the electronic network of practice, did not participate equally in the
production of the public good and that interactions were characterized by a
pattern of generalized reciprocity as opposed to one of a dyadic nature.
This finding is in line with work by Ekeh (1974) and Fulk et al. (1996).
Second, adapting Wenger’s (1998) categories of participation, we found
that the overall structure of the electronic network of practice was
characterized by a group of individuals who formed the critical mass that
was primarily responsible for the production of the public good. The
critical mass was then surrounded by a periphery of individuals who
participated to a lower degree. However, when we applied the
core/periphery measure that was developed for communities of practice, we
found that the electronic network was not characterized by a central core
of individuals closely connected to each other as in a community of
practice. Second, we performed a component analysis, which revealed that
this electronic network of practice was characterized by only one
component and not a set of subsets, indicating that this electronic network
did not have multiple cliques. Thus, while our findings are compatible
with critical mass theory (Oliver et al., 1985), we find that Wenger’s model
of participation categories needs to be amended when applied to electronic
Additionally, we examined the characteristics of the network members in

terms of heterogeneity of resources and interests. We found a high degree
of overall heterogeneity among the members, and in particular we found
that those who responded to a higher degree to others, and thus contributed
to the production of the public good, had a higher degree of resources
(professional and network expertise) and a higher level of interests (sole
practitioners and professional and social motives). These findings are
compatible with the work of Olson (1965), Hardin (1982), and Oliver et al.
(1985) who postulate that the heterogeneity of resources and interests of a
population affects collective action. One of the more interesting findings is
that which is in line with Hardin’s argument, namely that individuals who
lack private alternatives are those who have a high interest level in the
production of the public good (1982). In this electronic network of
practice, individuals who responded more often were more likely to be sole
DISCUSSION 197
practitioners and thus were less likely to have private alternatives for
professional discussion than a lawyer in a partnership law firm. The
electronic network of practice provided them then with important social
space for the development of their practice knowledge.
Applying our thinking on collective action and public goods back to the
community of practice study in Article 1, it is an interesting exercise to
consider whether knowledge within this community of practice exhibits
characteristics of a public good, even though this was not the express
purpose of our research in this study. First, we find that knowledge is by
its nature non-rival. Production of a community of practice’s knowledge
requires the same amount of input regardless of the number of community
members who will use it, and the use of this knowledge does not diminish
the availability of it to other members of the community. However,
previous research does not support the argument that a community of
practice’s knowledge is non-excludable to all members (Lave & Wenger,
1991). In a community of practice, knowledge entails a tacit component
that is learned through legitimate peripheral participation without being
made explicit. Due to the nature of interactions primarily being dyadic
within communities of practice, this knowledge is shared through one
individual interacting with another in face-to-face interactions. Thus, only
in the smallest of communities of practice could knowledge potentially be
non-excludable due to the physical restraints regarding the number of
dyadic interactions and relationships any one individual may have.
Additionally, as discussed, degrees of participation within a community of
practice are jointly determined by the members, with some individuals
moving towards the core while others are unable to move despite their
desire to do so. Thus, only individuals who are full members of a
community of practice have potential access to all the community’s
knowledge.
Building on our results of the community of practice study in Article 1, we

may then test the non-excludability of knowledge. If knowledge were non-
excludable in a community of practice, then the density of a community of
practice should be 100%, indicating that all individuals are interacting with
all the others. In this manner, individuals could not be excluded, in
principle, from the knowledge produced by others in the community.
However, when the density is less than 100%, individuals are not directly
tied to all the others and as such those acting as intermediaries can choose
198 CHAPTER SEVEN
to restrict the flow of knowledge, thus making the knowledge excludable to

others. Applying this thinking to the patterns of interaction within the
Technical Department, the construction project department that fulfilled the
structural properties of a community of practice, we find that the density is
only 27.5%, a far cry from 100%, despite the department comprising fewer
than 15 individuals. Thus, we find suggestive evidence that knowledge is
not non-excludable in this community of practice and thus is less of a
public good in a community of practice than in an electronic network of
practice.
Summary. Based on the above findings, we can now make some

conclusions regarding the examination of the structural properties of
networks of practice using social network analysis. First, our results
suggest that a common set of structural properties cannot be applied across
the board to all types of networks of practice. This in turn implies that a
generic set of social network measures cannot be applied to all networks of
practice. Rather, the relevant structural properties and corresponding social
network measures of a particular type of network of practice appear to be
dependent on the primary communication channels used by the network of
practice. The more a network of practice depends on electronic
communication channels such as bulletin boards or listservs in which all
interactions are visible to all network members, the more connected
members are. Thus, the network is denser and the distance between
members decreases in terms of simple structural interactions, implying that
the network’s knowledge increasingly develops characteristics of a public
good.
The one structural property that we do find to be common across the two
networks of practice studied is the core/periphery property, thus providing
support for community of practice theory (Lave & Wenger, 1991; Wenger,
1998) and inter-organizational network of practice theory (e.g., Schott,
1988). In both studies, we found a clear core surrounded by peripheral
members, with the core individuals exhibiting a higher degree of
professional and network expertise than the peripheral members who
surrounded them. Thus, our findings suggest that these core individuals are
influential in sharing knowledge with all members and in teaching novices
“how to get the job done”. However, as mentioned, we did still find a
difference in this structural property across the networks of practice
DISCUSSION 199
studied. Individuals within the core are not tied closely together in the
electronic network of practice as they are in the community of practice.
Finally, we find support for viewing knowledge as a public good and the
application of collective action to electronic networks of practice as
suggested by (Fulk et al., 1996; Monge et al., 1998). In addition, we find
partial support for applying these theories to other types of networks of
practice such as communities of practice.
These findings regarding structural properties thus suggest that the patterns
and dynamics of interactions and knowledge sharing among members vary
across the different types of networks of practice. In addition, these
findings have practical implications. To be more specific, our findings
have several implications for one of the central debates of networks of
practice as to whether they can be constructed and managed by
management. For example, our findings regarding the relationship between
the heterogeneity of resources and interests of individuals and critical mass
required for the creation of the public good suggest that the likelihood that
a network of practice develops and is sustained is dependent upon the
underlying constellation of individuals in terms of their resources and
interests. Thus, this implies that management may be able to successfully
support the development and sustainability of a network of practice by
ensuring a high degree of variance among the related resources and
interests of individual network members. We return to both the theoretical
and practical implications from these findings in the next section on
implications after we discuss our findings from Research Purpose 2.
7.1.2 Research Purpose 2: Individual Performance

The second research purpose is to investigate the relationship between
individual performance. As we found in the literature review, there are
few empirical studies that focus directly on this relationship. In addition,
the literature on the various networks of practice adopts the point of
departure that the individual has already decided to participate in the
network of practice in focus, and the general assumption is that the
individual aspires to be a full member of that network. However, as
previous research has indicated, individuals can make discretionary choices
regarding with whom they interact (Allen, 1977) and share knowledge
200 CHAPTER SEVEN
(Andrews & Delahaye, 2000), thus affecting the degree to which they
participate in various networks of practice.
To address this purpose, we conducted four studies in which we developed

a series of hypotheses grounded in the network of practice, knowledge-
based view of the firm, and social network literatures that predicted the
relationships between participation in various networks and individual
performance. Each of the studies contained a slightly different model;
however, they all investigated participation in various networks of practice
with individual performance as the dependent variable. The reason these
models differed are two-fold. First, the models were adapted to the
research site and the particular research questions of the study. Second,
with each study, we advanced our thinking based on both our previous
findings and on the work of other researchers. For example, the first Icon
study (Article 3) serves primarily as an exploratory study while the second
Icon study (Article 6) presents a considerably more extensive model.
Additionally, there are significant differences between the samples of
respondents. Cap Gemini respondents are all programmers located in the
Nordic countries while the respondents at Icon Medialab in the first study
come from a wide variety of functional backgrounds and were located in
nine offices spread across Europe and in the second study come from 28
offices across Europe, the United States, Asia, and Australasia. We discuss
and synthesize the findings from these four studies below.
Results from these four studies clearly indicate that individuals participate
to varying degrees in different types of networks of practice. Some
individuals participate only in communities of practice, while others
participate to a high degree in both inter-organizational electronic networks
of practice and communities of practice. We also find that an individual’s
level of participation in various types of networks of practice is related to
his or her individual self-reported performance. While there is some
variation in the findings among the studies, some clear trends do emerge.
The first relates to participation in communities of practice.
7.1.2.1 Communities of Practice

Community of practice theory suggests that members participate in the
sharing and transfer of valuable individual and group tacit knowledge, and
thus we would expect that individuals who participate to a high degree in a
DISCUSSION 201
community of practice would exhibit a higher degree of efficient

performance. The theory of weak ties (Granovetter, 1973) further suggests
that the knowledge in communities of practice is largely redundant due to
their closely-knit structures, providing little additional information over
what an individual may already know. Thus, a high degree of participation
and knowledge sharing in a community of practice would most likely
impede the ability of an individual to develop new and creative ideas,
resulting in a negative relationship between community of practice
participation and creative performance. We find mixed support for these
relationships. The research at Cap Gemini revealed that individuals who
participate to a higher degree in communities of practice report lower levels
of creativity; however, we found a different set of results at Icon Medialab
in Article 6. Here we found that individuals who participated to a higher
degree in communities of practice had both a higher degree of efficient and
creative performance. While the relationship to efficient performance is in
line with our a priori expectations, we did not expect to find a positive
relationship between community of practice participation and creative
performance at Icon.
One explanation for this finding at Icon may be due to the differences
between the two research sites. In the Cap Gemini study, individuals in the
same physical site were from one functional discipline, software
programming. Thus, when individuals discussed with others in their
communities of practice, knowledge was more likely to be redundant since
they shared the same general functional competence. However, due to the
organization of Icon as described above, the knowledge within
communities of practice at this company may have been less redundant. At
Icon, individuals within one physical location comprised several functional
disciplines, e.g., management consulting, programming, design, human-
computer interaction, etc., with individuals from a variety of functions
generally sitting next to each other in open environments. As a result,
communities of practice at Icon were more likely to incorporate more than
one functional competence. Since flexible integration involves integrating
existing knowledge in new patterns, the ability to achieve flexible
integration should increase when the number of different areas of
functional knowledge to be integrated increases due to a higher number of
potential combinations. Thus, one explanation for the difference in our
results could be that at Icon the more an individual participates in
communities of practice, the more likely they include individuals from a
202 CHAPTER SEVEN
variety of functional areas, thus potentially decreasing the level of

knowledge redundancy, which then positively impacts creative
performance.
7.1.2.2 Intra-organizational Distributed Networks of Practice

Looking at participation in intra-organizational distributed networks of
practice, the literature suggests that individuals in these networks share the
same organizational language and code of behavior and are faced with
similar issues related to their knowledge tasks, supporting integrative
efficiency and thus general performance (Brown & Duguid, 1991; Wenger,
1998). In addition, the combination and recombination of firm-specific
knowledge that is physically dispersed across the organization may
facilitate integrative flexibility and thus creative performance since these
acquaintances are more likely than community of practice members to have
important knowledge that is non-redundant (Granovetter, 1973). However,
due to less frequent patterns of interaction and lower intensities of social
pressure found in distributed networks of practice, members may be less
willing or committed to exchange knowledge without some type of return
(Blau, 1964). We would expect then that the relationship between
participation in intra-organizational distributed networks of practice and
individual performance would be mediated by internal knowledge
exchange. We found support for this at Cap Gemini where the relationship
between participation in intra-organizational distributed networks of
practice and performance was mediated by internal knowledge exchange.
In the second study at Icon, we further developed our model to include a
measure of network of practice centrality. Basing our argument on
previous social network research, we argued that an individual’s centrality
in a firm’s networks of practice mediates the relationship between
knowledge exchange and creative performance. The results from the Icon
study support these relationships.
These results then suggest that individuals who participate in a reciprocal

exchange of knowledge and help within intra-organizational distributed
networks of practice are more likely to become central individuals within
the firm’s networks of practice. These central individuals play the role of
brokers (Wenger, 1998) and, as mentioned previously, we may compare
them to boundary spanners in the technology transfer literature (Tushman
& Scanlan, 1981). Theory suggests that a central or broker position in the
DISCUSSION 203
networks of practice of a firm is jointly determined by the other members

of the networks of practice and the individual (Lave & Wenger, 1991). In
other words, just because an individual wishes to become a central
individual does not necessarily guarantee the individual such participation
status. Our results are in line with this, suggesting that individuals also
need to behave reciprocally and provide knowledge in exchange to other
members of their networks of practice to achieve central status.
Additionally, through their central position and collaboration with others in
distributed networks of practice, brokers are able to gather knowledge and
advice more efficiently and effectively from areas across the firm to fit
their own local needs than individuals in less central positions, thus
positively impacting individual performance.
While the findings regarding performance from the Cap Gemini and Icon
studies do differ somewhat, they do support the relationship between
individual performance. Through these distributed networks of practice,
individuals gain access to knowledge that may be reconfigured efficiently
to fit their local needs, resulting in more efficient performance (Cap
Gemini only), or they may gain access to new knowledge and innovative
ideas that they integrate with their own, resulting in more creative
performance (both Cap Gemini and Icon).
One explanation for the different results relating to efficient performance

may be due to the type of knowledge within the different samples. Cap
Gemini programmers are working primarily with software codes and as a
result may share a somewhat universal language, thus facilitating their
ability to communicate and share knowledge more efficiently across
organizational boundaries. Previous research on electronic networks of
practice of programmers has revealed that a large portion of
communication within these networks involves the transmission of blocks
of code (Wasko & Faraj, 1999). Furthermore, interviews in the first study
at Icon reported here indicated that blocks of code gathered by individuals
through their participation in distributed and electronic networks of practice
could easily be “cut and pasted” into their local work. As a result, we may
surmise that knowledge gathered by the Cap Gemini programmers from
distant locations through participation in intra-organizational networks of
practice may be easily absorbed locally and thus, positively impact efficient
performance.
204 CHAPTER SEVEN
Individuals at Icon, on the other hand, are working with a variety of

functional types of knowledge, e.g., human-computer interface, art design,
programming, etc., and these different functional types of knowledge may
be more local than universal. Explanations provided in interviews at Icon
were that the usage of the internet by end-users differed across countries,
thus limiting the degree to which front-end solutions developed in one
office could be efficiently “cut and pasted” between offices in different
countries. Rather, at Icon individuals needed to spend time and effort
translating and absorbing knowledge accessed through participation in
intra-organizational distributed networks of practice to fit their local
situations. Thus, we would expect that participation in intra-organizational
distributed networks of practice would not positively impact an individual’s
efficient performance.
Regarding creative performance at both Icon and Cap Gemini, we found

support that participation in intra-organizational distributed networks of
practice provides individuals with access to unique or non-redundant
knowledge, thus facilitating their ability to create new knowledge through
flexible knowledge integration and thereby increasing their ability to
achieve creative performance.
7.1.2.3 Inter-organizational Distributed Networks of Practice

Again we find differing results; however, they all support a positive
relationship between participation in inter-organizational networks of
practice and individual performance. Extending the model developed in
the first Icon study (Article 3), we introduced the measures of internal and
external knowledge exchange in the Cap Gemini study based on research in
the technology transfer literature (Article 5), arguing that external
knowledge exchange mediates the relationship between participation in
inter-organizational distributed networks of practice and performance.
However, contrary to expectations, we found that participation in inter-
organizational distributed networks of practice did not impact an
individual’s performance through external knowledge exchange. Rather
our findings suggest that external network participation and knowledge
exchange affect creative performance indirectly through influencing
internal knowledge exchange. One explanation may be that individuals
involved in external trading are performing gatekeeper activities similar to
those investigated in the research conducted in R&D operations by Allen
DISCUSSION 205
(1977) and others. Thus, these individuals may not be applying this
externally gained knowledge to their own tasks, which would affect their
own performance, as much as they are transmitting it to others within the
organization who have a specific need for this knowledge. A second
explanation may be that the knowledge acquired outside the firm may be so
novel that it must be combined with internal knowledge in order to be
applied in the firm’s context. This explanation is further supported by the
findings at Icon in which individuals who reported higher degrees of
creative performance were those who gathered knowledge through
participation in inter-organizational electronic communities and then
discussed these ideas with members of their communities of practice.
In the second Icon study, as mentioned above, we further developed our

model by incorporating the measure of network of practice centrality based
on the social network literature. We argued that the relationship between
network of practice participation and creative performance would be
further mediated by network of practice centrality. We found support for
this relationship, and interestingly, we also found a direct negative
relationship between participation in inter-organizational networks of
practice and centrality at Icon Medialab. Thus, those individuals who
interacted and shared knowledge with others outside the firm to a high
degree but did not integrate this knowledge with firm knowledge through
internal knowledge exchange remained on the periphery of the firm’s
networks and by implication achieved lower levels of creative
performance.
7.1.2.4 Electronic Networks of Practice

In the first Icon study (Article 3), we found a direct positive relationship
between participation in inter-organizational electronic networks of
practice and creative performance, however, a direct negative relationship
to efficient performance for the technically oriented individuals. As
discussed in Article 3, interviews provided several explanations for this
negative relationship, such as norms of reciprocity, the nature of
knowledge being shared, and individual professional motivations.
Based on previous research (Thibaut & Kelley, 1959; Blau, 1964;

Orlikowski, 1996), we extended our first Icon research and predicted that
participation in both internal and external electronic networks of practice
206 CHAPTER SEVEN
would be mediated by knowledge exchange. Our findings at Cap Gemini

support internal knowledge exchange in intra-organizational electronic
networks, and thus are compatible with previous research that norms of
reciprocity are critical for sustaining knowledge exchange in electronic
discussion networks (Lakhani & von Hippel, 2000; Wasko & Faraj, 2000).
However, contrary to expectations, we found no relationship between
participation in inter-organizational electronic networks and external
knowledge exchange. Thus, it seems that individuals are more likely to
engage in knowledge exchange with others through electronic networks
with whom they have a common bond, such as organizational membership.
We further discuss these differences in the following section on theoretical
implications.
7.1.2.5 Differences across Task Groups

Finally, the Icon studies revealed that there were differences in the general
patterns of participation in the various types of networks of practice and
their relationship to individual performance depending upon the task being
performed by the individual. For example, in the first Icon study (Article
3), we found that the technically oriented individuals made greater use of
external knowledge sources, such as electronic networks of practice, than
their commercially oriented counterparts, especially when solving difficult
problems. And as mentioned, a high degree of participation in inter-
organizational electronic networks of practice was negatively related to on-
time performance whereas this relationship was not found for the entire
sample.
In the second Icon study (Article 6), we found different patterns across the
three task groups in the relationships between participation in the various
networks of practice and performance. In particular, the System and
Software Development Group (SSW) was in strong contrast to the other
two groups of individuals, the Design Group and the Commercial and
Support Group, since SSW individuals exhibited a higher degree of
participation in external networks of practice. For example, the Design
Group exhibited a direct positive relationship between participation in a
community of practice and creative and efficient performance while the
SSW Group exhibited no direct relationship between these. Additionally,
while a negative relationship between participation in inter-organizational
networks of practice (both distributed and electronic) and network of
DISCUSSION 207
practice centrality was found for the entire sample as a whole, this
relationship was positive for the SSW Core.
As mentioned previously, one explanation for these findings may be linked

to the nature of the knowledge of the different groups. The pace of change
may be so fast within the system and software field that individuals must
rely to a high degree on the use of external knowledge sources to keep pace
with development. Knowledge within the firm may quickly become out of
date and the reuse of this “old” knowledge may hamper the ability to
develop and implement new solutions and processes. In addition, a second
explanation may be due to the more universal nature of system and
software knowledge; external knowledge used by the SSW Group may not
need to be translated before it is combined with internal knowledge through
knowledge exchange to the same degree as the external knowledge
gathered by the Commercial and Support Group and Design Group.
7.1.2.6 Participation in Networks of Practice and Firm Performance

Finally, while not the express purpose, the study of the three high-
technology multinationals investigating how MNCs manage knowledge
flows in their global R&D operations (Article 7) provides tentative
evidence of a positive relationship between participation of individuals in a
firm’s intra-organizational distributed network of practice and firm
performance.
Summary. Our empirical studies reveal a significant relationship between

an individual’s participation in various types of networks of practice and
individual performance. On the one hand, we find that efficient
performance has a direct, positive relationship to participation in
communities of practice, yet too much participation in communities of
practice comprising members sharing the same functional expertise may
lead to a lower degree of creative performance. On the other hand,
inter-organizational networks of practice, be they electronic or distributed,
has a positive relationship to creative performance. However, our results
also reveal that knowledge exchange and centrality are important mediators
in the relationship between participation in networks of practice and
creative performance. Further investigation of these performance
relationships revealed significant differences between groups of individuals
208 CHAPTER SEVEN
based on their tasks, indicating that the dynamics of knowledge sharing

within the various networks of practice is contingent upon the underlying
practice knowledge. We discuss the practical and theoretical implications
of these findings in the next section.
7.2 Theoretical Implications
As stated in Chapter One, the overarching goal of this thesis is to improve

our understanding of networks of practice from a business firm’s
perspective, and our approach has been to bring together a number of
related theories that can inform our understanding of both individual-level
behaviors and network-level activities as opposed to the organizational
level. In this manner, we may then propose implications relevant to the
network of practice literature. Additionally, our intention is to apply our
thinking and findings back to the level of the firm so that we may then
contribute to the extant literature on the knowledge-based view of the firm.
We discuss these two areas in turn.
7.2.1 Networks of Practice

Chapters Two and Three, in which the literature and the relevant empirical
studies were reviewed, provided a point of departure for this thesis, leaving
us with an understanding of the current “state of affairs” of the network of
practice field. To structure our discussion, we developed a network of
practice matrix, mapping the various types of networks onto this matrix
based on the primary communication channels and nature of the network of
practice. The review of empirical studies conducted to date on the various
kinds of networks of practice revealed that in general there is a dearth of
studies within all areas of the network of practice matrix. Additionally, we
found that the majority of studies explicitly focused on networks of practice
have investigated the cognitive aspects of communities of practice. Thus,
we identified structural properties of networks of practice as well as the
relationship between networks of practice and performance as two
significant gaps in the research. A third gap is that there are an extremely
limited number of studies focusing on intra-organizational distributed
networks of practice and electronic networks of practice. Thus, this review
exposes that “what we think we know” is considerably more than “what we
know” regarding networks of practice.
DISCUSSION 209
The set of seven empirical studies presented in this thesis helps to fill these
research gaps as well as support previous thinking regarding networks of
practice. Regarding the structural dimensions of networks of practice, our
studies show that the synthesis of social network concepts and methods
with the network of practice literature greatly facilitates the ability to
uncover these “invisible” networks. Thus, this research opens the door for
additional theorizing and empirical studies on the structural properties of
various networks of practice. For example, we may use structural
properties to help detect and analyze networks of practice, to track their
development over time, or to measure their relationship to performance.
These findings regarding structural properties also reveal that the patterns
and dynamics of interactions and knowledge sharing among members vary
across the different types of networks of practice. Due to the close
relationship between structural and cognitive dimensions, we suggest that
this then implies that the cognitive dimensions also vary across networks of
practice. For example, results from our study of the electronic network of
practice (Article 2) call into question whether or not participation in an
electronic network of practice is jointly determined as it is in a community
of practice. Due to the nature of the media in an electronic network of
practice, individuals can post and respond to others to the degree they
desire regardless of whether other individuals are interested in interacting
in a relationship with the individual posting the messages. Other
differences revealed relate to the reach and reciprocity of various networks.
Comparing the two studies (Articles 1 and 2), we find that reach is much
more extensive and that reciprocity is of a general nature in the electronic
network of practice. In addition, we provide support for applying theories
of collective action and public goods to electronic networks of practice and
suggest that these may also be applied to other types of network of practice
to facilitate our understanding of them.
In the second set of studies (Articles 3 to 7), we focus on the research gap
concerning the relationship between participation in various types of
networks of practice and performance. Our results suggest that this
relationship is not only contingent upon the strength of the tie but also upon
the redundancy of the knowledge in the network at hand. Our findings
suggest that the strong ties of communities of practice have a positive
impact on members’ efficient performance. However, the redundancy of
the knowledge in terms of functional competencies in the community of
210 CHAPTER SEVEN
practice impacts members’ creative performance. Thus, the more a

community of practice is characterized by a diversity of functional
competencies, the more likely that the community of practice is able to
develop more creative solutions through the recombination of these diverse
competencies. However, for communities of practice characterized by the
same functional competency, there is greater likelihood that this
community of practice may turn into a competency trap or core rigidity
(Levitt & March, 1988; Leonard-Barton, 1992) unless members of this
community of practice also participate in distributed or electronic networks
of practice in which they may access non-redundant knowledge.
Thus, not only are our findings compatible with Granovetter’s theory of
strong ties, but they also provide suggestive evidence for social resources
theory (e.g., Lin, Ensel, & Vaughn, 1981). Social resources theory focuses
on the nature of the embedded resources in a network and argues that it is
not the weakness of the tie per se that conveys advantage, rather it is the
likelihood that the tie reaches someone with the required resource. We
modify this to our findings by arguing that it is not the strength of the tie
per se within the network that affects creative performance, but it is the
composition of the network in terms of the diversity of knowledge
competencies that it comprises.
Additionally, we find some indication that the degree to which the

network’s practice knowledge is of a more universal nature and less
embedded in a local practice is an important aspect in the dynamics of
networks of practice. Our results suggest that practice knowledge that
generally is of a more universal nature, such as software programming,
may be shared more effectively by individuals participating in distributed
and electronic networks of practice than knowledge of a less universal
nature, such as the design knowledge of the art directors at Icon. This is
evidenced by the finding that the technically oriented individuals used
external electronic media including electronic networks of practice to a
higher degree than their commercially or support-oriented counterparts at
Icon, especially when solving difficult problems. In addition, further
evidence is found at Cap Gemini where programmers who participated to a
higher degree in distributed and electronic networks of practice exhibited a
higher degree of efficient performance. These findings are in line with
previous research within the technology transfer literature in which it was
hypothesized that individuals conducting tasks of a more universal nature
DISCUSSION 211
could communicate more effectively across organizational boundaries than

individuals conducting tasks of a less universal nature (Allen, Tushman, &
Lee, 1979).
Our findings also imply that other cognitive dimensions such as norms of
reciprocity and trust differ across the various types of networks of practice.
For example, we do not find a relationship between participation in inter-
organizational electronic networks of practice and external knowledge
exchange. However, we do find a relationship between participation in
inter-organizational distributed networks of practice and external
knowledge exchange. One explanation may be found by looking at the
relationship between the ability to establish reciprocal exchange and
various communication channels. Building reciprocal relationships with
individuals in inter-organizational electronic networks may be more
difficult since members generally have not met each other face-to-face and
have little social influence over one another due to the voluntary and
anonymous nature of the exchange. In addition, when reciprocity occurs in
these networks it is typically of a general and not a dyadic nature (Kollock,
1999). In order for an individual to give to the network, there must be a
level of trust across the network members that ensures other network
members will “pay back” when requested. Achieving this level of trust and
reciprocity may require a more complex process than one-on-one, face-to-
face relationships.
Our empirical studies also indicate that it may be easier to build trust and
achieve a norm of reciprocity in intra-organizational electronic networks
than in inter-organizational ones. This may be because individuals within
these networks have a common organizational tie and are thus working for
the greater good of the company (Constant et al., 1996). However, there
are other aspects to consider. Intra-organizational networks may be more
stable in terms of participation, membership, and identification of
participants. It is also possible that individuals are not as anonymous as
they are in inter-organizational networks. Finally, misbehavior in an intra-
organizational electronic network may be more easily “punished” and carry
tangible deterrents, while positive behaviors may be rewarded through
increases in status and reputation in the organization. As a result, intra-
organizational electronic networks may be able to better control their
boundaries and member behavior, resulting in more effective knowledge
flows.
212 CHAPTER SEVEN
Our findings also support taking a differentiated view of networks of

practice over a unitary one. The rationale for taking a unitary view of
networks of practice may be meaningful in that networks of practice may
be described as structures that foster knowledge sharing and learning
within business firms. However, we find that imposing unitary views on
networks of practice masks possible heterogeneity along two dimensions.
Through our ability to contrast the findings from the series of seven
empirical studies in this thesis, we have uncovered differences in networks
of practice along two dimensions: 1) the type of the network and 2) the
underlying practice. First, in general if we look at the findings related to
performance and structure, we find significant differences across the
various types of networks of practice. For example, knowledge sharing in
communities of practice is positively related to general performance while
knowledge sharing in intra-organizational distributed networks of practice
is positively related to creative performance. Additionally, communities of
practice are characterized by a more restricted reach and dyadic reciprocity
than electronic networks of practice. Secondly, by conducting the analyses
on the different task groups within the Icon studies, we find that networks
of practice may differ based on the knowledge that is the basis for the
network’s practice. For example, as we see in table 7.1, the drivers of
network of practice centrality are different for each of the three task groups
in the second Icon study (Article 6), with participation in external networks
of practice as a positive driver of network of practice centrality for the
System and Software Group but a negative driver for the Commercial and
Support Group. Additionally, in the first Icon study (Article 3) we found
that the technically oriented individuals (e.g., system architects and
software programmers) attached greater importance to their contacts in
external networks, often turning first to individuals in inter-organizational
networks of practice for help. Our analysis also revealed that the
technically oriented employees rated external sources of knowledge as
more helpful than the non-technically oriented employees did. One of the
explanations for these differences offered in the qualitative data was that
external sources provided a broader source of expertise and access to the
latest thinking within the field, which was especially important for
technically oriented employees due to the fast pace of technological
change. Thus, these findings suggest that unitary views assuming common
characteristics across different networks of practice risk losing their power
in explaining interesting outcomes.
DISCUSSION 213
Table 7.1 Drivers of Network of Practice Centrality at Icon (Article 6)

Task Group
Drivers of System Commercial

Network of Practice Centrality and Software and Support Design
1. Use of internal codified sources, e.g.,

intranet Positive
2. Participation in intra-organizational
Positive
distributed networks of practice
3. Internal knowledge exchange

Positive Positive
4. Participation in inter-organizational
distributed and electronic networks of Positive Negative
practice
However, taking a differentiated view of networks of practice leads to a

different way of thinking about these emergent networks. For example,
instead of taking a unitary view of networks of practice in firms such as
Icon Medialab that comprise numerous competence groups, a more refined
approach would be to study networks of practice within each competence
group. In this manner, variations related to structure, performance, and
cognitive aspects may be uncovered. Furthermore and as mentioned below
under practical implications, this thinking implies that generic efforts to
support networks of practice across competence groups in a firm may be
unsuccessful due to their inability to meet the specific needs of the network
of practice within each group.
Additionally, while we have chosen not to investigate the antecedents of

network of practice participation decisions in this thesis, we may further
speculate that the above differences in patterns may be due to underlying
differences in individual personal characteristics. In other words, choices
to participate in different networks of practice may partly depend on an
individual’s personality type or even their education or previous training.
For example, at the risk of broadly generalizing, we may speculate that
individuals such as those who conduct commercial tasks in sales may be
more likely to choose to communicate to a higher degree via face-to-face
interactions with individuals outside the firm whom they know due to sales
training or perhaps even a more extrovert personality while software
214 CHAPTER SEVEN
programmers may be more likely to choose to work with anonymous others

in inter-organizational electronic networks of practice due to programming
training or a more introvert personality. Research within the technology
transfer literature supports this idea. For example, based on their research
in a U.S. military R&D lab, Taylor & Utterback (1975) are of the opinion
that a person becomes a gatekeeper partly based on personal inclination.
Furthermore, Taylor (1975) speculated that individuals in R&D working on
basic and applied research tasks are more likely to communicate with
others outside their project team than those working on product
development tasks since researchers are generally trained to seek help from
others as opposed to attempting to solve problems based on their own
resources, as technicians generally are trained to do.
Our findings also contribute to several of the debates within the network of
practice literature. A review of the studies to date as well as the empirical
findings from the studies in this thesis provide support for the frequency of
face-to-face interactions as being a determining factor in the type of
network that emerges. Thus, as indicated in our definition of communities
of practice, we argue that these networks only emerge within groups of
individuals whose primary communication channel is frequent face-to-face
interactions and as a result are more likely to develop among groups of
individuals who are co-located.
A second debate is whether innovation occurs within communities of

practice or at their interstices. While the level of analysis is at the
individual, our findings provide suggestive evidence for both sides of the
argument. We propose that the likelihood of innovation occurring within
communities of practice depends upon the degree of the heterogeneity of
the backgrounds of the community of practice’s members. We also find
suggestive evidence for the second argument that innovation occurs at the
interstices of communities of practice through brokers as evidenced by the
creative performance of individuals participating in networks of practice
that span internal and external organizational boundaries. However, we
also find suggestive evidence that communities of practice may evolve into
core rigidities and competency traps (Levitt & March, 1988; Leonard-
Barton, 1992) as evidenced by the negative relationship between creative
performance and participation in communities of practice.
DISCUSSION 215
A final contribution to the network of practice literature is that by

conducting a set of seven empirical studies and including them in this
thesis helps to advance our understanding of networks of practice to a
greater degree than if these studies were conducted as separate works by
different authors. Having conducted these complementary studies
personally helps to analyze and compare the findings based on deeper
insights, thus providing a more complete view of networks of practice. In
order to facilitate future work on networks of practice, in table 7.1 we
propose our own definitions of the various networks of practice based on
the understanding that we have developed in this thesis. Additionally, we
present a summary of the characteristics of the various networks of practice
in table 7.2.
Table 7.2 Definitions of the Various Types of Networks of Practice

Network of Practice Definition
Community of practice An emergent group of a limited number of individuals
contextually bound through close face-to-face interactions
in the pursuit of a common enterprise
Intra-organizational An emergent group of an unlimited number of dispersed

distributed network of organizational members working on similar tasks using a
practice similar competence who are generally acquainted with one
another through dyadic relationships
Intra-organizational An emergent group of an unlimited number of dispersed

electronic network of organizational members working on similar tasks using a
practice similar competence whose communication channel is
purely internet-based
Inter-organizational An emergent group of an unlimited number of dispersed

distributed network of individuals regardless of organizational affiliation working
practice on similar tasks using a similar competence who are
generally acquainted with one another through dyadic
relationships
Inter-organizational An emergent group of an unlimited number of dispersed

electronic network of individuals regardless of organizational affiliation working
practice on similar tasks using a similar competence whose
communication channel is purely internet-based
216 CHAPTER SEVEN
Table 7.3 Characteristics of the Various Types of Networks of Practice

Characteristic Communities of Intra-organizational Inter-organizational Electronic Networks of
Practice Networks of Practice Networks of Practice Practice *
Location of members Co-located Both co-located and Distributed Distributed
distributed
Primary Face-to-face Mixed face to face, Mixed face to face, Computer-mediated,
communication electronic and other electronic and other threaded messages
channels forms of communication forms of communication
such as news letters such as news letters
Membership Restricted, locally Somewhat restricted Somewhat restricted Open
bounded since bounded by since bounded by
personal relationships personal relationships
Size / reach Limited to same time, No limitations but No limitations but No limitations
same place interaction generally smaller due to generally smaller due to
dependency on personal dependency on personal
relationships relationships
Participation Limited to co-location Open and based on Open and based on Open and based on shared
and shared practice shared interest in shared interest in interest in practice
practice practice
Nature of interaction Jointly determined, Partly jointly Partly jointly Individually determined,
visible social and status determined, partly determined, partly filtered social and status
cues filtered social and status filtered social and status cues
cues cues
Identity High degree of shared Medium to medium-high Medium to medium-low Low degree of shared
identity, language, degree of shared identity, degree of shared identity, identity, language, norms,
norms, values language, norms, values language, norms, values values
Restraint on behavior High Medium Medium Low
DISCUSSION 217
Characteristic Communities of Intra-organizational Inter-organizational Electronic Networks of

Practice Networks of Practice Networks of Practice Practice *
Strength of member Strong ties based on Weak ties based on Weak ties based on Weak ties based on online
ties personal relationships affiliation in the practice affiliation in the practice interaction
and shared experiences,
and a common history
Knowledge High redundancy and Medium to medium-high Medium to medium-low Low redundancy and low
high level of shared tacit redundancy and medium redundancy and medium level of shared tacit
knowledge. to medium-high level of to medium-low level of knowledge. Public good
Characteristics of private shared tacit knowledge. shared tacit knowledge. characteristics.
good Both private and public Both private and public
good characteristics. good characteristics
Structural High degree of Medium degree of Medium degree of High degree of
characteristics connectedness. connectedness. connectedness. connectedness.
Individuals of critical Individuals of critical Individuals of critical Individuals of critical
mass tied to one another. mass tied to one another. mass tied to one another. mass not tied to one
Direct reciprocity Both direct and Both direct and another. Generalized
generalized reciprocity generalized reciprocity reciprocity
Performance High degree of efficient Low degree of efficient Low degree of efficient Low degree of efficient
but generally low degree but high degree of but high degree of but high degree of creative
of creative creative creative
Note: We have combined intra-organizational and inter-organizational electronic networks of practice in one category. Clearly there are
differences between the two, e.g., an intra-organizational electronic network of practice’s membership is limited to the total number of
employees in an organization. However, the primary purpose of this table is to show the differences between the four major types of
networks of practice, hence we have combined the types of electronic networks of practice.
218 CHAPTER SEVEN
7.2.2 The Knowledge-based View of the Firm

Observed by Grant (1996), theories of the firm are conceptualizations of
business enterprises that explain and predict structure and behavior. While our
research has not explicitly been aimed at understanding firm theories, as
mentioned in Chapter One, our intent in this thesis is to apply our thinking and
empirical findings on networks of practice on both individual-level behaviors
and network-level activities back to the level of the firm such that we may then
contribute to the extant literature on the knowledge-based view of the firm.
7.2.2.1 Grant’s Theory of Knowledge Integration

As mentioned under Research Purpose 2, a common theme in empirical research
on the knowledge-based view of the firm focuses on knowledge integration. As
Grant argues (1996a), the firm’s primary task is to integrate the specialized
knowledge of multiple individuals – through various coordination mechanisms
such as rules, sequencing and routines. In addition, a high degree of team
interdependence involving group problem-solving and decision-making is
required. Competitive advantage results then from how effective firms are in
integrating the specialized knowledge of their members, and Grant proposes that
this effectiveness depends upon the efficiency, the scope, and the flexibility of
knowledge integration. Our research is compatible with Grant’s arguments of
efficient and flexible knowledge integration since our findings indicate that
participation and knowledge access in various types of networks of practice
impact individual performance in different ways. For example, we found at Cap
Gemini that those individuals who participated to a high degree in communities
of practice exhibited a low degree of creative performance; however, those who
participated in intra-organizational distributed and electronic networks of
practice exhibited a high degree of creative performance.
However, as mentioned in Chapter Four, Grant’s theory focuses primarily on

issues of coordination (structuring to enhance the effectiveness of knowledge
integration), without referring to issues of “cooperation”. As such, this theory
leaves out a key component by assuming that people are willing to share
knowledge openly and freely without expecting anything in return if provided
with the structures/opportunities. Our empirical studies suggest otherwise. Both
our quantitative and qualitative findings indicate that this view of the firm as a
knowledge integrator should be further developed by incorporating a dimension
of cooperation. Norms of reciprocity and expectations of returns for knowledge
sharing appear to be key factors of participation and knowledge access in all
types of networks of practice. Thus, in order for an individual to access
DISCUSSION 219
knowledge from members in a network of practice, our results indicate that he or

she must be willing to provide knowledge in return.
7.2.2.2 The Firm as a Social Community and Community of Communities

In addition to the above and in connection with the growing interest in networks
of practice, there is also a significant body of literature arguing that a distinctive
feature of the firm is its capacity for enabling the transfer of tacit knowledge
(Kogut & Zander, 1992, 1996). Kogut & Zander (1992:384) write, “In our
view, the central competitive dimension of what firms know how to do is to
create and transfer knowledge efficiently within an organizational context.”
These authors further argue that the firm should be understood as a social
community, writing “We suggest that organizations are social communities in
which individual and social expertise is transformed into economically useful
products and services by the application of a set of higher-order organizing
principles. Firms exist because they provide a social community of voluntaristic
action structured by organizing principles that are not reducible to individuals”
(1992:384). The main idea is that firms are communities within which the
transfer and combination of knowledge are facilitated through a shared common
stock of knowledge, shared coding schemes, and a shared language. Costs of
communication are lower between the firm’s members due to a shared identity,
which also results in shared organizing principles reflected in the firm’s explicit
and tacit rules of coordination and which influences the direction of search and
learning (Kogut & Zander, 1996). Thus, firms are more efficient vehicles than
the market in their transfer of tacit knowledge and in their ability to coordinate
the combination of varieties of functional expertise.
Further proponents of this knowledge-based view of the firm are Brown &
Duguid (1991, 1998), who argue that the firm should be seen as a community of
communities. They write, “most formal organizations are not single
communities of practice, but, rather, hybrid groups of overlapping and
interdependent communities” (1998:97). While interest in these views of the
firm continues to grow, with the exception of studies such as that by Zander &
Kogut (1995), there are interestingly few empirical studies investigating these
views. However, our findings are both compatible with and suggest further
areas for development within this view.
Support. First, our findings are compatible with the view of the firm as a social
community and a community of communities (figure 7.1). At the local level, we
find that individuals are members of communities of practice, with a high degree
of participation in these communities leading to a high degree of efficient and, in
some circumstances, creative performance. Individuals within these local
220 CHAPTER SEVEN
communities may then also be members of intra-organizational distributed and

electronic networks of practice. Individuals participating to a high degree in
these networks serve as brokers, bridging local communities of practice through
exchanging, transferring, and translating knowledge between them. Theory
suggests that the transfer of knowledge between these individuals is facilitated
by a shared common stock of knowledge, language, and norms, and that new
knowledge is created through the combination of local knowledge with less
redundant distant knowledge within the firm. Our studies provide suggestive
evidence of this since we found that those individuals who participated to a
higher degree in intra-organizational distributed networks of practice exhibited a
higher degree of creative performance.
In addition, Brown & Duguid (2000) view the firm as being interconnected with
other firms through their members’ participation in inter-organizational
networks of practice that tie together individuals from a variety of external
organizations. Our results are compatible since we find that individuals
regularly tap into external sources of knowledge through participation in inter-
organizational networks of practice to get their work done. In the first Icon
study, we even found that the production-oriented technically oriented people
made greater use of external sources of knowledge than their commercially
oriented counterparts, especially when solving difficult problems. Thus, in
knowledge-intensive firms such as Icon and Cap Gemini, employees are able to
communicate across external organizational boundaries with others working on
similar problems and as a result, access new knowledge and ideas.
Within these views of the firm, proponents argue that one source of innovation
lies in the interface between a firm and its environment (Brown & Duguid,
1991) and that the creation of new knowledge occurs through the combination of
novel external knowledge with internal knowledge (Kogut & Zander, 1992). In
line with this, we find in our quantitative results that the access of external
knowledge through participation in inter-organizational networks of practice
does not have a direct impact on an individual’s creative performance. Rather, it
affects performance indirectly through its influence on internal knowledge
exchange. Thus, our findings suggest that individuals combine externally
gathered knowledge with their own as well as with knowledge obtained within
the firm through participation in intra-organizational networks of practice.
Additionally, as our qualitative findings from the first Icon study (Article 3)
suggest, this external knowledge may need to be translated in order to adapt and
combine it for the firm’s specific use.
DISCUSSION 221
Figure 7.1 The Firm as a Social Community and Community of

Communities
IONP: Inter-organizational network of practice

IANP: Intra-organizational network of practice
CP: Community of practice
Firm
Environment
IONP
Unit A
CP
IONP
CP
IONP
Core Individuals/Brokers
IANP IANP
CP
CP
IONP
NP
Unit B
Finally, the view of the firm as a social community also argues that performance
differences among firms partly arise due to the ability of firms to transfer
knowledge within their boundaries due to shared coding schemes, shared values,
shared identity, and higher organizing principles (Kogut & Zander, 1992). We
found suggestive evidence for this claim in our final study that focused on the
R&D operations of three multinationals. In this study, we observed that the
highest performing firm, Hewlett-Packard, is the one that exhibited a higher
degree of shared identity by members across the firm’s geographically distant
operations as well as higher levels of knowledge sharing between members of
intra-organizational distributed networks of practice.
Areas for Development. In addition, our findings also indicate some areas for
further development in the field. First, at the individual level, our studies
indicate that the issue of an individual’s membership in the firm should be
considered. Within networks of practice, one of the primary themes in both the
research on networks of practice to date as well as in our empirical studies is the
degree to which an individual is a member of a particular network of practice. If
you recall, Wenger (1998) proposes that there are different levels of
222 CHAPTER SEVEN
participation within a community of practice: full participation (insider),

peripherality, marginality, and full non-participation (outsider). In order for an
individual to obtain access to the community and its knowledge, it is necessary
for the individual to become an “insider”. This occurs through a process of
legitimization through mutual engagement, collaboration, and storytelling,
during which the individual learns the values and the informal and technical
language of the community, while most importantly, how to function as a
community member (Lave & Wenger, 1991). Insiders are full members at the
core of the community, having fully mastered the community language as well
as the community rules and how to use them. They know such things as who is
involved with what activities and what is needed to become a full member of the
community, and they are able to perform the practice in a more effective means
than those in the periphery.
However, the traditional perspective on organizational membership in the

literature views membership as a binary state59. Yet if the firm is to be viewed
as a social community, then we should be able to apply this network of practice
thinking regarding membership as well to the firm. The argument then would be
that individuals are “members” of a firm to differing degrees, as a function of
their time in the firm, their participation in mutual engagement and collaboration
with other firm members, their ability to master the language and norms of the
firm, their degree of shared identity, and even their desire to become a full
member of the firm. Individuals who are insiders in the firm due to their having
fully mastered the firm’s practice, language, and rules and how to use them
would be found at the core of the firm. We may then hypothesize that the
degree to which an individual is a member of the firm and in the core of the
firm’s entire network of internal networks of practice will be associated with a
higher degree of individual performance. What this implicitly argues then is that
59
Building on Weber, several researchers see individuals as being either members or non-
members of formal organizations (Scott, 1998). Aldrich (1979: 221) writes, “The minimal
defining characteristic of a formal organization is the distinction made between members and
nonmembers, with an organization existing to the extent that entry into and exit out of the
organization are limited. Some persons are admitted, while others are excluded.” Following
Weber (1947: 140), organizational membership is then defined in the following way: “A party
to a closed social relationship will be called a ‘member’.” This perspective holds that
organizational authorities control the entry and exit into and from an organization and set the
conditions for member entry and exit. In addition, authorities control “wages or salaries,
hours of work, amount of work expected, and the allocation of a member’s organizational
time” (Aldrich, 1979: 222). Thus, in this view, individuals are either a member or a non-
member. Individuals employed by a firm are members since the firm’s authorities have
admitted them under certain entry conditions, and those who are not employed are non-
members.
DISCUSSION 223
full firm members will have mastered the practice of the firm through mutual
engagement and collaboration (as predicted by theory) and will benefit through
superior performance, whereas individuals who are less firm-like in their
behavior and do not collaborate with others in the firm will have inferior
performance. If such a hypothesis is not supported, then we have evidence that
mutual engagement, collaboration, and membership are not valuable to
performance, which would throw into doubt the overlying argument that we see
the firm as a social community.
While it may seem strange to claim that some people are more “members” of a
firm than others, not only is this view consistent with network of practice theory,
but it is also entirely consistent with the concept of fuzzy logic. Fuzzy set
theory, introduced by Lofti Zadeh in the 1960s, implements classes or groupings
of data whose boundaries are not sharply defined, i.e., fuzzy. In this manner,
fuzzy set theory deliberately blurs the rules of logic that insist that categories are
unambiguously applied and that there is an “excluded middle” between
something being and not being the case. The benefit of applying fuzzy
techniques is the strength in solving real-world problems, which inevitably
entail some degree of imprecision (Battelle, 1997). While we do not propose
that there is a direct correspondence between the use of the concept in
mathematics and our application here, we do feel that this concept may improve
our understanding of the membership of individuals in firms. In addition, this is
not the first time that this term has been used in the management literature.
Hagström (2000) used this term to describe the networks of a firm’s
relationships that fall between market-type and legal firm relationships.
The results from our extensive social network study of the entire body of
employees at Icon in 26 offices in 16 countries (n=1698, Article 6) are
compatible with this view of the firm as a social community with individuals
exhibiting different degrees of firm “membership”. First, we find that
individuals who participate to a higher degree in communities of practice within
Icon’s subsidiaries and thus are full members of the local practice exhibit a
higher degree of individual performance. We also find that individuals who
participate in mutual engagement, collaboration, and knowledge exchange with
others throughout the firm to a higher degree are those who are in the core of the
firm’s internal networks of practice (network of practice centrality), and these
individuals also have a higher degree of creative performance. Finally, we find
a direct negative relationship between participation in inter-organizational
networks of practice and centrality. Thus, those individuals who merely access
knowledge externally to a higher degree but do not combine this knowledge
with that of others through participation in the firm’s networks of practice
224 CHAPTER SEVEN
remain on the outskirts of the firm and by implication achieve lower levels of
creative performance.
Another area for further consideration that is related to firm membership regards
the individual motivations of firm members. Kogut & Zander view the firm as a
“social community of voluntaristic action”, arguing that the “assumption of the
selfish motives of individuals resulting in shirking and dishonesty is not a
necessary premise in our argument” (1992: 384). Thus, if we understand the
argument correctly, their view is based upon the assumption that individuals
within a firm are generally willing to collaborate and share their knowledge with
one another. Applying our thinking from theories of public goods and collective
action, the implicit underlying assumption then is that the firm should be viewed
as a collective with the knowledge produced by the firm as a public good.
However, as our literature review on electronic networks of practice revealed,
previous studies indicate that individuals make choices regarding their
willingness to share with others based on the expected payoffs. Research on
electronic networks of practice, an organization that is perhaps closer to being a
collective with knowledge as a public good than a traditional business firm,
provides evidence that individuals are motivated not only by collective interest
(e.g., interest in advancing the community) but also by self-interest (e.g.,
enjoyment, increased reputation) (Lakhani & von Hippel, 2000; Wasko & Faraj,
2002). Our study of an electronic network of practice (Article 2) reveals that
those individuals who are the most willing to participate through sharing their
knowledge and thus in the core of the network are motivated by self-interest
since they are concerned with enhancing their own reputations. However, those
individuals who ask others for help are motivated by collective interest since
they view their participation as a means to sustain the network.
Applying this thinking to the firm, we would then expect that firm members
make choices regarding their participation and willingness to share with others
within the firm based on expected individual and collective payoffs. However,
since individuals also may participate in inter-organizational networks of
practice, the indication is that individuals weigh payoffs from internal
participation with payoffs from external participation. In some cases, as
indicated in the first Icon study (Article 3), external payoffs may even outweigh
internal payoffs. For example, in the qualitative data we found that individuals
chose to leak the firm’s proprietary knowledge to individuals in other firms in
return for a self-benefit of increased reputation or a collective benefit of
advancing the professional community. As a result and in line with our
argument above, some individuals may then choose to be lesser members of the
firm than others due to higher expected returns from external participation.
DISCUSSION 225
Finally, in addition to individual motivations to participation and knowledge

sharing, community of practice theory would argue that participation in the firm
is jointly determined. Thus, although some individuals desire to become full
members of the firm through choices to participate and share their knowledge
with others within the firm, they may not be allowed to do so by other
individuals within the firm. Thus, the individuals found in the periphery in the
second Icon study (Article 6) may not all be there by choice. Some of these
individuals may be there because they are not allowed into the firm by other
firm members. Thus, the only available sources for help may be external ones.
Following from the above, our findings also suggest then that the boundaries of
the firm in knowledge-intensive companies such as Icon and Cap Gemini are
less rigid than traditional theory would imply, with individuals tapping regularly
into informal external sources of knowledge. While the knowledge-based view
argues that a firm’s boundaries provide a demarcation in identity with members
of a firm attaching meaning to their firm membership and having a shared
identity (Kogut, 2000), we may further suggest that this may be expanded. First,
drawing from the network of practice literature, this shared identity implies that
the boundaries of the firm are not only determined by the willingness of an
individual to share with others in the firm but also by the ability, or rather the
inability, of the individual to use externally accessed knowledge within the
firm’s local context and practice. In other words, individuals develop a shared
identity due to their willingness to interact and share with others in the firm and
as a result are less able to apply external knowledge than internal knowledge in
their work. However, building on the above discussion of varying degrees of
membership and community of practice theory, we suggest that a firm may be
characterized by a core of individuals with a higher degree of shared identity
surrounded by rings of individuals with decreasing degrees of shared identity.
The more an individual chooses or is “restricted” to participate in external
networks of practice while participating less in the firm’s networks of practice,
the more likely this individual may have a lower degree of shared firm identity
but a higher degree of shared external identity. In some cases, an individual
may even have a higher degree of shared external identity than firm identity.
Seen in this light, we propose then that a firm’s boundaries in addition to its
members are fuzzy as well.
This speculation regarding firm boundaries and membership also has

implications when discussing the hybrid form of organization. Most arguments
concerning the hybrid organization think of the hybrid in terms of contractual
relationships between two parties that are usually at the organization level, e.g.,
joint venture, alliance between two companies, etc. (Williamson, 1991).
226 CHAPTER SEVEN
However, our results indicate that hybrids can occur at the individual level as
well, with some individuals falling somewhere between full-scale firm members
and independent contractors. Thus, it could be hypothesized that this individual
behavior makes some “hybrid” organizational form likely - in which individuals
are still members of firms, but they also exhibit certain market-like behaviors.
A second area for development within the knowledge-based view of the firm
relates to our findings regarding the structural properties of networks of practice.
If we view the firm as a social community and community of communities, then
building on our findings from Research Purpose 1, we may develop a set of
appropriate structural properties. For example, we would expect to see within
the firm as a whole a similar constellation to that of various types of networks of
practice: a core of full members surrounded by rings of peripheral members and
perhaps even non-participants. Other properties involving measures such as
density, graph-theoretic distance, and connectedness could then be developed to
help determine the degree to which knowledge sharing is occurring across the
firm or language, values, and norms are potentially shared. These structural
properties could then facilitate further theorizing within the knowledge-based
view of the firm or facilitate a means to compare firms on various dimensions.
7.2.2.3 Absorptive Capacity

On a final note, our results are also in line with absorptive capacity theory by
Cohen & Levinthal (1990). Absorptive capacity has been defined as the ability
of a firm to recognize the value of new, external knowledge; assimilate it; and
apply it to commercial ends. The authors argue that a firm’s absorptive capacity
is dependent upon the absorptive capacities of its individual members and is not
only a factor of the firm’s direct interface with the environment, but it also is a
factor of the firm’s ability to transfer knowledge from the point of entry to
throughout the firm. Furthermore, in order to assimilate and use new
knowledge, individuals and the firm need to have prior related knowledge.
Our findings further suggest that absorptive capacity and performance are
enhanced by the combination of new knowledge with existing knowledge that
crosses intra-organizational boundaries through individuals participating in
knowledge exchange within intra-organizational distributed networks of
practice. Our individual performance findings further indicate that knowledge
coming from outside the firm cannot be easily transferred into and applied to
any immediate solution inside the firm. Rather, external knowledge accessed by
individuals through participation in inter-organizational networks of practice
must be translated to the firm’s local context and practice through a high degree
DISCUSSION 227
of personal interaction and knowledge exchange through internal networks of

practice.
Finally, we further suggest that we may borrow from the work on absorptive
capacity by proposing that ties between individuals can be characterized by their
absorptiveness. We feel that the dimension of strength does not adequately
reflect the ability to absorb knowledge through a tie. Thus, we propose that the
characteristic of absorptiveness is based on two dimensions of ties: strength and
the degree of shared related knowledge. In other words, if a tie between
individuals is characterized by a high degree of absorptiveness, then the
individuals may easily assimilate and use knowledge shared through the tie due
to a high degree of shared related knowledge as well as a high degree of
strength. An example of this is individuals who are members of the same
community of practice. However, a medium to medium-high level of
absorptiveness may still be achieved through ties that are weak, contrary to what
social network theory might suggest. For example, in an inter-organizational
electronic network of practice for programmers working with C++, a member
may have a weak tie with other members in the network of practice since he
participates primarily through lurking. However, the individual could still be
able to rather effectively absorb knowledge accessed in this electronic network
of practice for use in his work due to a high degree of absorptiveness stemming
from extensive prior experience in C++. At the same time, this individual may
have a series of strong ties with other individuals who have a low level of prior
related knowledge, e.g., colleagues from another department with whom one
eats lunch, thus the ability to absorb relevant knowledge through these ties is
low to medium-low. Lastly, ties that are both weak as well as characterized by a
low degree of prior related knowledge have a low degree of absorptiveness, e.g.,
colleagues in another department located in another location. We illustrate our
thinking in Figure 7.2.
In summary, our research makes several contributions not only to the network of
practice literature but also to the knowledge-based view of the firm. In addition
to these theoretical implications, there is also a set of practical implications,
which is the subject of the next section.
228 CHAPTER SEVEN
Figure 7.2 Absorptiveness of Ties
High
Medium to
Medium- High
Degree of High
Shared Related
Knowledge
Low to
Low
Medium-Low
Low
Weak Strong
Strength of Tie
7.3 Implications for Practice
7.3.1 Knowledge Management

One of the primary implications of the results from our research is for the field
of knowledge management. Our findings indicate that organizations concerned
with knowledge management may need to rethink their knowledge management
strategies. First, what is to be made of the high use of external knowledge
sources? To date, knowledge management systems have focused on leveraging
knowledge within the organization. In some instances, these systems may also
include other organizations (sometimes referred to as extranets), yet these other
organizations fall within the realm of the organization’s formal task
environment. However, as our studies reveal, informal external sources are used
highly by individuals in addition to having a positive influence on creative
performance and in some cases a negative influence on efficient performance.
A challenge then for management is whether knowledge management systems
should be developed that facilitate the use of external sources. At Icon,
interviews with technically oriented individuals group reveal that these
individuals prefer going outside the organization to external electronic
communities to asking others within Icon for help. Not only could more
answers be found to their questions, but also the speed with which answers were
given was much faster than using internal knowledge sources.
Second, what do we make of the result that different patterns of knowledge

sourcing and participation in networks of practice are associated with different
performance outcomes? With the rapid development of the ease of use of the
DISCUSSION 229
internet (e.g., smart agents, more specialized discussion forums) and the
increasing ability of individuals to use the internet to communicate with others
in their external networks of practice, this media is expected to become a much
more helpful knowledge source. While the use of external sources and
participation in inter-organizational networks of practice has a positive
relationship to creative performance, we find, however, that it is participation in
communities of practice that leads to superior efficient performance. The
question then becomes how to balance the use of external networks of practice
with internal ones to ensure a productive ratio of creativity to efficient
performance. One suggestion is that knowledge management systems could be
aimed at facilitating a balance between efficient and creative performance that
matches a company’s competitive strategy. As we find here, knowledge
integration patterns differ depending upon which type of performance is the
objective. In some organizations, a focus on efficient performance through
systems that promote local communication may be the objective while in others
a focus on creative performance through systems that promote the development
of inter-organizational networks of practice may be the objective.
In addition, this research shows that implementing generic knowledge

management strategies across an organization may not prove successful. As we
observed, the technically oriented individuals at Icon differed significantly both
in their use of various knowledge sources and the relationship to individual
performance and centrality. These differences imply that knowledge
management systems tailored to each group of employees based on its practice
knowledge may be more successful. For example, for groups such as the
System and Software Group, a system focused on the use of external knowledge
sources as well as internal ones may be a better use of a firm’s resources than a
system that only focuses on internal ones.
This research also indicates that internal knowledge management systems may
be more successful if they focus on linking individuals together as opposed to
focusing on knowledge repositories. Results are in line with previous research,
suggesting that individuals prefer to communicate with others when searching
for knowledge. These results then suggest an important new use of internet-
based communication technologies to support knowledge management. Rather
than using technology to replace traditional knowledge management techniques,
such as creating document repositories, management may need to think of non-
traditional ways to leverage these new technologies for improved knowledge
flows within and across the firm, by leveraging networks that support the
exchange of advice and ideas between individuals.
230 CHAPTER SEVEN
Our findings also suggest some practical implications for the development and
maintenance of electronic networks of practice. First, our study indicates that
electronic networks of practice do not need equal member participation, but
rather they can be sustained through the collective actions of a small percentage
of members who form a critical mass. This critical mass is able to provide the
public good through generalized exchange of advice and solutions. These
individuals are often concerned with enhancing their reputations in the network,
thus technology that supports identifiers of individuals may be more likely to
succeed than systems where participation is anonymous. In addition, we found
that those most likely to develop the critical mass were tenured experts in their
area, but may not have easy access to other interested individuals. Thus, unlike
communities of practice that require face-to-face interaction, electronic
networks of practice transcend traditional barriers to knowledge exchange
through the creation of knowledge as a communal public good, available to all
members of the collective.
Finally, our research findings suggest that intra-organizational distributed

networks of practice are a valuable complement to traditional face-to-face
communities of practice. Learning in a distributed or electronic network of
practice appears to be on a broader, more general level (i.e., learning about new
areas and topics within the practice) while learning in a community of practice
may be of a deeper, more specific type (i.e., learning how to apply the new area
to one’s specific task). This distinction may be characterized by a “T” format as
shown in figure 7.3. We argue that in order to improve an organization’s ability
to efficiently integrate knowledge while simultaneously creating new knowledge
through flexible knowledge integration, organizations should focus on
sponsoring both traditional communities of practice and distributed and
electronic networks of practice as well as stimulating the interaction between
the two. Through the active participation of organizational members in both
communities and networks, these organizational forms become integrated, thus
resulting in the exchange and recombination of existing knowledge to create
competitive advantage. In addition, the integration of these two may serve as a
defense mechanism against the “not-invented-here” syndrome, core rigidities,
and competency traps.
DISCUSSION 231
Figure 7.3 T-based Network of Practice Learning
Network of Practice
Broad-based learning
Community of Practice
In-depth learning
7.3.2 Participation in Inter-organizational Networks of Practice

Just as individuals have a certain degree of commitment to their organizations,
they also have a degree of commitment to their profession or occupation as
several researchers have noted (e.g., Saxenian, 1996; Brown & Duguid, 2001).
In some professions, the degree of commitment to the profession can be so
strong that the norms of the profession even transcend the norms of the
organizations that employ the individuals. Members of professions can be
separated by great distances and still see themselves as part of the same
professional group. Academics have long been examples of individuals with a
strong degree of commitment to their profession (Pickering & King, 1995).
The implication from our results is that individuals may then hold multiple
identities, as Kogut & Zander (1996) posit. On the one hand, individuals belong
to their firm, while on the other, they may belong to a professional network
outside the firm, as evidenced through the high participation in inter-
organizational networks of practice. Thus, an individual may be faced with
competing allegiances and conflicting objectives. This can be best described
through our qualitative findings from the first study at Icon in which it was
found that these conflicting identities were strong within Icon, primarily among
the programmers. While programmers were inspired by management to make
Icon the world’s best company, they were, however, pressured by their external
internet communities to produce the latest “cool” solution. In addition,
programmers were under social pressure from their external networks to help
232 CHAPTER SEVEN
fellow members solve their difficult problems, often attempting to “show off” in
front of the others. This was found to lead to conflicting goals for the
programmers: best company vs. best function.
One result of this participation in inter-organizational networks of practice is

that individuals may spend too much time “working for” their external network.
The first Icon study revealed that individuals who spent time working with
others in electronic communities were more likely to have a poor level of
efficient performance. Interviews with some of these individuals revealed that
they often were so busy helping others outside of Icon or striving to create
elegant or “bleeding edge” solutions to impress others in their external networks
of practice that they were unable to focus on finishing their own tasks according
to management’s objectives. Furthermore, it appeared that these individuals had
considerable “power” over management. This power resulted from
management’s inability to understand in detail what their employees were doing
since they were unable to keep up with the rapid pace of technological
development. Thus, management often did not know whether employees were
working on necessary value-adding activities or were spending time trying to
impress their peers. One manager summarized this situation with reference to
the software programmers, “Programmers take us (management) hostage. We
never know whether they’re working on extra bells and whistles to impress their
buddies or whether it’s really a value-adding activity for the customer.” For
some individuals, the payoffs may be greater from developing the “latest and
coolest” solution than from completing their work on time and to their
supervisor’s requests.
The second challenge is more of a threat to a firm’s competitive advantage – the

trading of proprietary know-how or information. The development of the
internet has greatly facilitated the degree by which proprietary information can
be traded between companies without management’s knowledge. In many
cases, management is completely unaware that its employees are engaging in
this behavior. The decision whether to transmit proprietary knowledge is placed
within the hands of the individual. If an individual is trying to enhance his or
her identity in the external network, then he or she may leak this proprietary
knowledge without receiving anything of value for the company in return.
However, the results of this research seem to provide contrary evidence, that
external knowledge exchange is beneficial for the firm, although indirectly
through the recombination with existing knowledge. This indirect relationship
also implies that while an individual may trade away “proprietary” knowledge,
the ability of a rival firm to turn this into an innovation lies in its ability to
integrate it with its own existing knowledge bases in the firm.
DISCUSSION 233
Third, while some organizations have attempted to stop the cross boundary flow
of knowledge within these inter-organizational network, these attempts may be
counterproductive (Brown & Duguid, 2001). Attempting to block participation
in inter-organizational networks of practice may only lead to increased loyalty to
the external network and decreased loyalty to the organization and potentially a
negative effect on individual performance. And as noted above, these flows are
two-way, generally characterized by an equal exchange of knowledge. Thus,
cutting off flows to outside of the firm will more than likely risk cutting off
flows into the firm (Saxenian, 1996; Brown & Duguid, 2001). Interviews at
Icon conducted considerably after data collection in the second study revealed as
well that after management focused on increasing the development of solutions
through internal knowledge gathering and reducing participation in external
networks of practice, the level of creative solutions and the ability to compete
dropped in addition to a number of high performers leaving the firm.
Finally, one implication of this research is that when management hires a

person, management is also “hiring” the employee’s external network as well.
Thus, management must consider the potential employee’s external network and
how active the individual is in this network. If the person is very active in this
external network, then the individual’s time may be spent on external activities.
As indicated here, individuals who participate to a high degree only in external
knowledge exchange and not in internal exchange, are less central in the firm
and may exhibit a lower degree of efficient performance. Another aspect to
consider is which individuals are included in the potential employee’s network
and what value these individuals may provide to the organization through
external knowledge exchange.
As a result, it can be suspected that individuals will most likely have to balance
their participation in the firm with their participation in external networks. In
some cases, individuals may feel that their relationship with their employer is
subordinate to their relationship with their outside network. They may feel that
they can always get another job through their network and they may take their
network with them no matter where they work. Thus, this leads us to question
whether the internet is creating a second professional environment external to
the company that now can compete with and displace the professional
environment of the company. The success of an individual may no longer be
tied solely to his or her performance within a company. In one sense we are
beginning to see this already in Silicon Valley where individuals claim that they
work for “the Valley” and not for any one particular company (Saxenian, 1996).
As these external professional groups grow, they may become potential rivals to
managerial control if the profession has a higher influence than the organization
234 CHAPTER SEVEN
on the individual. This may occur when the individual has a higher degree of
commitment or loyalty to the profession than to his or her organization. In
addition, the professional group to which an individual belongs may have a
strong code of norms and values that management has difficulty in influencing
(Pickering & King, 1995).
7.3.3 Brokers
This research has also provided evidence that management should pay
considerable attention to developing an understanding of who the brokers are in
the organization since they are of considerable importance to the firm. Brokers
are influential in the future of the firm since they are one of the primary
determinants in the direction of the firm’s knowledge development. Due to their
central position in the firm’s intra-organizational networks, brokers wield power
over resources. As a result, they influence the organization’s knowledge flows
between firm units and the knowledge sourcing processes of other individuals.
Brokers provide their own knowledge or provide pointers to relevant experts in
the firm or even outside the firm when others from across the organization come
to them for help. Additionally, they are critical in determining which external
knowledge is combined with which internal knowledge due to participation in
knowledge exchange. While most individuals throughout the firm participate in
some kind of external knowledge exchange, it is easier for brokers to trade away
the firm’s more valuable knowledge since they have greater access to the firm’s
more valuable resources and critical knowledge due to their position. Decisions
to trade and the ability to obtain valuable external knowledge are based on their
own judgment. Finally, brokers may also influence the degree to which other
individuals become true firm members. In summary, since knowledge is argued
to be the most valuable resource of the firm (Kogut & Zander, 1992; Grant,
1996a,b, Nahapiet & Ghoshal, 1998), brokers are critical to the firm’s
knowledge integration processes and the ability to create sustainable competitive
advantage.
7.3.4 Achieving the Balance

Thus, one challenge for management is to achieve the appropriate balance of
participation in inter-organizational and intra-organizational networks of
practice by firm members to facilitate the optimal mix of efficiency and
creativity. While we have not specifically studied how this might be achieved,
we provide here some areas for consideration by management. One suggestion
is that management might focus on aligning the balance of participation in
internal and external networks of practice with the company’s competitive
DISCUSSION 235
strategy. For example, if the firm is pursuing more of a knowledge creation than
a knowledge reuse strategy (Hansen, Nohria, & Tierney, 1999), then
encouragement of participation in external networks of practice over internal
ones may facilitate creativity and access to the latest solutions. However, if the
firm’s strategic focus is on knowledge reuse, then too much external
participation may lead to an inefficient use of resources. Additionally, as we
noted above, taking a generic view of networks of practice throughout the
competence groups in the firm may lead to poor results, indicating that the
appropriate balance may also need to be reconsidered for each competence
group.
Furthermore, if there is one takeaway from this research for managers, it is that
efforts to “control” participation and knowledge sharing in various networks of
practice by firm members will more than likely be fruitless or even counter-
productive. For example, in the studies in the literature review as well as those
performed for this thesis, we found that individuals participated in external
networks of practice and consciously traded company secrets with others in
competitive firms despite it being against company policy. Thus, one means of
achieving the appropriate participation levels in the different networks of
practice may be for management to interview potential employees regarding
their problem-solving and knowledge-gathering behaviors during recruiting and
to select those whose problem-solving behaviors match the needs of the firm.
Other means may include management clearly communicating the firm’s
knowledge management strategy (e.g., knowledge creation vs. knowledge reuse)
and knowledge-sourcing strategies and aligning human resource policies such as
evaluation and remuneration practices to reflect the desired balance. Finally,
various socialization efforts may also help in two ways: 1) to increase
participation and knowledge sharing in communities of practice and intra-
organizational distributed networks of practice and 2) to increase employee
loyalty and commitment to the firm such that individuals make the appropriate
decisions when trading knowledge across a firm’s boundaries.
7.3.5 The Future Firm

One area for final speculation is what the firm of the future will look like.
While we are still in the first years of internet use, one of the main questions that
this research raises is what will organizations look like thirty years from now
when a generation who has never known life without the internet will have
entered the work force. These individuals are learning to use internet-based
communication tools in a completely different manner than today’s users in
firms. One of the main complications for many of today’s internet users in firms
236 CHAPTER SEVEN
is that they have difficulty believing that they can build trust-based relationships
with people with whom they communicate over the internet and have not met
face-to-face. This affects their ability to then work together with others purely
over the internet and thus their ability to exchange the more valuable tacit
knowledge with one another. Many of today’s users are stuck in old ways of
learning to trust others based on physical presence – body language, appearance,
tone of voice, etc. However, as individuals increase their use of the internet,
they may learn how to trust others based on other cues, for example the
timeliness of others’ responses, the ways in which messages are formulated, etc.
This will then affect their ability to work together with others over the internet
and thus the degree to which they choose to participate in both intra-
organizational and inter-organizational electronic networks of practice and thus
their firm membership.
This speculation then has potential ramifications for the firm as we know it
today. Will the dominant organizational form then become one of “fluid and
temporary networks” of independent contractors rather than traditional firms?
Today this organizational form is already dominant in certain industries such as
the film industry. Will this form continue to grow in certain industries, but
never gain ground in certain others that will continue to be dominated by face-
to-face traditional firms? These questions provide an exciting area for future
research, which is the subject of the next section.
7.4 Limitations and Suggestions for Future Research
Despite the strengths of the comprehensiveness of this thesis, we should also

note its limitations. On a more general note, first, we have taken primarily a
structural approach in the studies and as such, we have generally not focused on
any of the cognitive aspects of networks of practice when conducting our
analyses. One potential research area then is the interaction between the
structural and cognitive dimensions of the various types of networks of practice.
Second, the studies relied on cross-sectional data collection. On one hand, this
methodology is appropriate for model testing. However, a cross-sectional
approach tells us little about how the social context develops over time, what
factors influence its development, and how motives and social context interact
and change over time. Thus an area of future research could examine network
dynamics at different points in time, as well as how individuals’ participation in
various networks changes over time.
DISCUSSION 237
Generalizability of the study findings is also an issue. It is difficult to assess

how representative the findings are for other organizations and networks of
practice than the ones examined here. Although the focus was on knowledge
workers exchanging knowledge, networks of practice within lesser knowledge-
intensive firms or consisting of scientists, researchers, or engineers as well as
non-business organizations may have different membership and exchange
dynamics. Thus, studies comparing various networks of practice and their
dynamics across a variety of organizations are suggested.
Additionally, in this research, our definition of knowledge and knowledge

sharing incorporated the whole scale from tacit to explicit knowledge. Nor did
we look at the strength of the relationships between individuals in the advice
networks. However, previous research suggests that at the unit level the
relationship between performance and the position in the firm is dependent upon
the type of knowledge being transferred as well as the strength of the tie
(Hansen, 1996). The work by Jarvenpaa & Staples indicates that a distinction
should be made between information and expertise and that the relationship
between organizational ownership and self-ownership of information and
expertise should be examined. Finally, as Bouty’s study revealed,
confidentiality is socially constructed and there are open secrets. Thus, future
research should incorporate these different dimensions of knowledge, tie
strength, knowledge ownership, and confidentiality.
On a more specific level, for the studies on the structural dimensions of a

community of practice and an electronic network of practice, we examined only
one specific type of network of practice, thus limiting our ability to generalize to
other similar networks of practice. For example, other types of electronic
community interactive technology exist such as chatrooms, and the use of
different media may affect electronic network of practice dynamics. In addition,
these studies relied on data that were primarily cross-sectional. Thus, we were
not able to investigate changes over time. Subsequent studies should be
longitudinal to understand network of practice lifecycles. Longitudinal analysis
should also look at the nature of interdependence of individuals’ decisions to
contribute to the public good in the case of electronic networks of practice. It
has also been argued that reciprocal interdependence and not sequential
interdependence characterizes interactive communication systems (Fulk et al.,
1996). However, it has yet to be tested empirically. In addition, studies should
be conducted on the other various types of networks of practice.
A final issue of interest and further research is the aspect of lurkers. In

electronic networks of practice, individuals may free-ride through lurking,
238 CHAPTER SEVEN
reading all messages to gain access to the network’s knowledge without ever
posting themselves. There is also the question of whether people who
continually ask questions, receive help from the electronic network of practice,
but never bother to help anyone else in the electronic network of practice are
free-riders. It can be argued that these individuals actually do contribute to the
public good because they stimulate a thought process by other participants or
provide support through side posting. However, this type of participation only
works if there is a critical mass of individuals who continue to respond to
postings. Finally, while a participant may be interested in contributing, if there
is low electronic network of practice activity, then he or she may feel that their
actions will not be reciprocated the next time they need help and thus, their time
spent helping is lost.
As for the studies investigating individual performance, this research only

examined individuals within two companies, thus limiting the generalizability of
our findings. Thus, further research should examine knowledge integration and
the relationship to individual performance across multiple organizations.
Second, data were collected at one point in time. Another limitation is our use
of self-reported survey measures for performance. Thus, future research should
include other performance data sources as well as the collection of data over
time to further establish the relationships between participation in various
networks of practice participation and performance. Additionally, we have
primarily investigated performance at the individual level. As we found in the
review of empirical studies, research looking at the relationship between
different types of networks of practice and their dynamics with performance at
different levels, such as the network or the organization, is needed. Finally, it
may be argued that our results regarding individual performance seem to
oversimplify the picture. As indicated above, we have not investigated the
antecedents regarding individual decisions to participate and share knowledge in
different networks of practice. Not only may personal characteristics influence
which individuals participate in which networks of practice, they may even
further influence outcomes once the choices are made, thus providing another
interesting area for research.
In terms of the knowledge-based view of the firm, we have uncovered several

areas for future research. First, Grant (1996a) proposes three dimensions of
effective knowledge integration. We investigate two of these: efficient and
flexible knowledge integration. However, due to the already complex nature of
this study, we did not investigate the third dimension, the scope of integration.
Clearly, it would be of interest to look at the effect of the scope of integration
and its relationship to individual performance. Second, research on the firm as a
DISCUSSION 239
social community and community of communities should examine the motives

of individuals and their choices regarding their participation and willingness to
share knowledge with others within the firm vs. others outside of the firm.
Another interesting research area is the investigation of differences in patterns
among individuals from different functional groups.
Although not discussed here, with respect to organization design, the traditional
view of organization design proposes that interactions with the environment take
place through formal inter-organizational relations or through formal boundary
spanning roles in the periphery of the organization. However, our results
indicate that individuals are embedded not only in intra-organizational networks
of practice but that they are also embedded in networks of practice that extend
across an organization’s external boundaries. Individuals regardless of position
and task rely to a high degree on these external networks of practice for advice
and knowledge in solving their everyday work tasks. In the process, individuals
exchange knowledge, accessing external knowledge and combining it with
internal knowledge. These findings suggest that the boundaries of the firm are
not as porous as traditional theory has suggested. Thus, concepts such as the
Thompsonian protected technical core surrounded by boundary spanning
individuals performing support activities such as salespeople may no longer
apply and should be investigated (Thompson, 1967).
On a final note, while studies to date on the structural properties of networks of

practice have been limited due to difficulties in collecting data, the ability to
conduct studies investigating the structural properties of networks through social
network analysis is improving dramatically due to advances in internet-based
communication. For example, Tyler, Wilkinson, and Huberman (2003) describe
in a working paper a methodology for the automatic identification of networks
of practice from email logs at HP labs. Using 185,773 logged emails during a
two-and-a-half month period between HP employees, they argue that their
method is effective in identifying both formal networks and networks of
practice. Thus, one exciting area for future research is to explore the ability to
use these new communication media as a tool to investigate various types of
7.5 Conclusions
In summary, the overarching goal of this thesis has been to improve our
understanding of networks of practice from a business firm’s perspective and in
particular to investigate issues of structure and performance. First, we
240 CHAPTER SEVEN
conducted an extensive literature review of the research conducted on the

various networks of practice to provide us with a current “state of affairs”.
Second, we conducted a series of seven complementary empirical studies that
addressed two research purposes based on the identified gaps of structure and
performance within the network of practice literature. Results from the studies
addressing the first research purpose provided the development and testing of
various network of practice structural properties and revealed that the two polar
forms of networks of practice have significantly different patterns of knowledge
flows, suggesting that the various types of networks of practice differ in terms of
their cognitive dimensions. The studies addressing the second research purpose
provide evidence that an individual’s performance is related to participation in
various types of networks of practice. Finally, this thesis provides a more
comprehensive view of networks of practice than if the studies were conducted
separately due to our ability to compare the various networks of practice on
various dimensions based on both the literature review as well as our empirical
studies. As a result, this thesis suggests taking a differentiated view of networks
of practice over a unitary one since imposing one view on networks of practice
masks possible heterogeneity along two dimensions: 1) the knowledge of the
practice and 2) the form of the network of practice.
In addition to contributions to the network of practice literature, we then applied

our findings to the extant knowledge-based view of the firm literature and
suggested areas for future development. In particular, the findings of the last
two studies, one based on an extensive social network study of a modern
multinational and the other based on case studies of three large, traditional
multinationals, are compatible with the view of the firm as a social community
and a community of communities in which individuals are fuzzy members.
Finally, we provided a discussion of our findings in terms of implications for
practice before leaving the reader with an understanding of where future
research could be conducted.
References
Abbott, A. 1993. The sociology of work and occupations. Annual Review of

Sociology, 19: 187-209.
Ackerman, M.S., Pipek, V., & Wulf, V. 2003. Sharing Expertise: Beyond
Knowledge Management. Cambridge: MIT Press.
Adler, E. & Haas, P. 1992. Conclusion: epistemic communities, world order,
and the creation of a reflective research programme. International
Organization, 46,1: 367-390.
Aiken, M. & Hage, J. 1968. Organizational interdependence and intra-
organizational structure. American Sociological Review, 33,6: 912-930.
Alavi, M. & Leidner, D.E, 1999. Knowledge management systems: Issues,
challenges and benefits. Communication of the Association for Information
Systems, 1: 1-28.
Aldrich, H. 1976. Resource dependence and interorganizational relations: Local
employment service offices and social services sector organizations.
Administration & Society, 7,4: 419-454.
Aldrich, H. 1979. Organizations & Environments. Englewood Cliffs: Prentice-
Hall.
Aldrich, H. & Herker, D. 1977. Boundary spanning roles and organization
structure. Academy of Management Review, April: 217-230.
Allen, R.C. 1983. Collective invention. Journal of Economic Behavior and
Organization, 4: 1-24.
Allen, T.J. 1970. Communication networks in R&D laboratories. R & D
Management, 1: 14-21.
Allen, T.J. 1977. Managing the Flow of Technology: Technology Transfer and
the Dissemination of Technological Information Within the R&D
Organization. Cambridge, MA: MIT Press.
Allen, T.J. & Cohen, S.I. 1969. Information flows in research and development
laboratories. Administrative Science Quarterly
Allen, T.J., Hyman, D.B. & Pinckney, D.L. 1983. Transferring technology to
the small manufacturing firm: A study of technology transfer in three
countries. Research Policy, 12: 199-211.
Allen, T.J., Tushman, M.L. & Lee, D.M.S. 1979. Technology transfer as a
function of position in the spectrum from research through development to
technical services. Academy of Management Journal, 22,4: 694-708.
Aloni, M. 1985. Patterns of information transfer among engineers and applied
scientists in complex organizations: A partial review. Scientometrics, 8, 5-6:
279-300.
242 REFERENCES
American Productivity & Quality Center (APQC). 2001. Building and

Sustaining Communities of Practice: Continuing Success in Knowledge
Management. Houston, Texas: AQPC.
Andrews, K.M. & Delahaye, B.L. 2000. Influences on knowledge processes in
organizational learning: The psychosocial filter. Journal of Management
Studies, 37,6: 797-810.
Appleyard, M. 1996. How does knowledge flow? Interfirm patterns in the
semiconductor industry. Strategic Management Journal, 17: 137-154.
Argyris, C. & Schön, D. 1978. Organizational Learning: A Theory of Action
Perspective. Menlo Park: Addison-Wesley.
Augoustinos, M. & Walker, I. 1995. Social Cognition: An Integrated
Introduction. London: Sage.
Bacharach, S. B., & Lawler, E. J. 1980. Power and Politics in Organizations.
San Francisco: Jossey-Bass.
Bagozzi, R.P. & J.R. Edwards. 1998. A general approach to representing
multifaceted personality constructs: Application to state self-esteem.
Structural Equation Modeling, 1: 35-67.
Baldwin N.S. & Rice R.E. 1997. Information-seeking behavior of securities
analysts: Individual and institutional influences, information sources and
channels, and outcomes. Journal of the American Society for Information
Science, 58, 8: 674-693.
Barley, S. 1986. Technology as an occasion for structuring: Evidence from
observations of CT scanners and the social order of radiology departments.
Administrative Science Quarterly, 31: 78-108.
Barley, S. 1996. Technicians in the workplace: Ethnographic evidence for
bringing work into organization studies. Administrative Science Quarterly,
41,3: 404-441.
Barnard, C.I. 1938. The Functions of the Executive. Cambridge, MA: Harvard
University Press.
Bartlett, C.A. & Ghoshal, S. 1989. Managing Across Borders: The
Transnational Solution. Cambridge: Harvard Business School Press.
Battelle Memorial Institute. 1997. Fuzzy Logic.
http://www.emsl.pnl.gov:2080/proj/neuron/fuzzy/what.html.
Bechky, B. 1999. Creating shared meaning across occupational communities:
An ethnographic study of a production floor. Presented at the Academy of
Management Conference, Chicago, IL.
Bechky, B. 2003. Sharing Meaning Across Occupational Communities: The
Transformation of Understanding on a Production Floor. Organization
Science, forthcoming.
Bender, T. 1982. Community and Social Change in America. New Brunswick,
NJ: Rutgers University Press.
Bentler, P.M. 1980. Multivariate analysis with latent variables: Causal
modeling. Annual Review of Psychology, 31: 419-456.
REFERENCES 243
Bentler, P.M. & Bonett, D.G. 1980. Significance tests and goodness of fit in
analysis of covariance structures. Psychological Bulletin, 88: 588-606.
Berger, P.L. & Luckman, T. 1966. The Social Construction of Reality. London:
Penguin Books.
Berkowitz, S.D. 1988. Afterword: Toward a formal structural sociology. In B.
Wellman & S.D. Berkowitz (eds.), Social Structures: A Network Approach.
Volume 15 in Contemporary Studies in Sociology, New York: Cambridge
University Press.
Birkinshaw, J. & Hagström, P. (eds.) 2000. The Flexible Firm. London: Oxford
University Press.
Blackler, F. 1993. Knowledge and the theory of Organizations: Organizations
as activity systems and the reframing of management. Journal of
Management Studies, 30,6: 864-884.
Blackler, F. 1995. Knowledge, knowledge work and organizations: An
overview and interpretation. Organization Studies, 16,6: 1021-1046.
Blau, P.M. 1964. Exchange and Power in Social Life. New York: Wiley.
Blau, P.M. 1973. The Organization of Academic Work. New York: John Wiley
& Sons.
Blau, P.M. 1977. Inequality and Heterogeneity: A Primitive Theory of Social
Structure. New York: Free Press.
Boisot, M.H. 1995. Information Space: A Framework for Learning in
Organizations, Institutions, and Culture. London: Routledge.
Boisot, M. & Griffiths, D. 1999. Possession is nine tenths of the law: Managing
a firm’s knowledge base in a regime of weak appropriability. Journal of
Technology Management, 17,6: 662-676.
Boland, R. J., Jr. & Tenkasi, R. V. 1995. Perspective making and perspective
taking in communities of knowing. Organization Science, 6, 4: 350-372.
Bommer, W.H., Johnson, J.L, Rich, G.A., Podsakoff, P.M., & MacKenzie, S.B.
1995. On the interchangeability of objective and subjective measures of
employee performance: A meta-analysis. Personnel Psychology, 48,3: 587.
Borgatti, S.P., & Everett, M.G. 1999. Models of core/periphery structures.
Social Networks, 21: 375-395.
Borgatti, S.P., Everett, M.G., & Freeman, L.C. 1999. UCINET 5.0 for Windows.
Analytic Technologies.
Borman, W.C. 1978. Exploring upper limits of reliability and validity in job
performance ratings. Journal of Applied Psychology, 63,2: 135-144.
Bourdieu, P. & Wacquant, L.J.D. 1992. An Invitation to Reflexive Sociology.
Chicago: Chicago University Press.
Bouty, I. 2000, Interpersonal and interaction influences on informal resource
exchanges between R&D researchers across organizational boundaries.
Academy of Management Journal, 43, 1: 50-65.
244 REFERENCES
Brass, D.J. 1985. Men’s and women’s networks: A study of interaction patterns
and influence in an organization. Academy of Management Journal¸28: 327-
343.
Brent, D. A. 1994. Information technology and the breakdown of "places" of
knowledge. Ejournal, 4, 4: online.
Brown, J.S. & Duguid, P. 1991. Organizational learning and communities of
practice. Organization Science, 2,1: 40-57.
Brown, J.S. & Duguid, P. 1993. Rethinking the border in design: An
exploration of central and peripheral relations in practice. In S. Yelavich
(Ed.) The Edge of the Millennium: An International Critique of Architecture,
Urban Planning, Product and Communication Design. New York: Whitney
Library of Design: 174-189.
Brown, J.S. & Duguid, P. 1998. Organizing knowledge. California Management
Review, 40,3: 90-111.
Brown, J.S. & Duguid, P. 2000. The Social Life of Information. Boston:
Harvard Business School Press.
Brown, J.S. & Duguid, B. 2001. Knowledge and organization: A social-practice
perspective. Organization Science, 12, 2: 198-213.
Burawoy, M. 1979. Manufacturing Consent: Changes in the Labor Process
under Monopoly Capitalism. Chicago: University of Chicago Press.
Burgoyne, J.G. & Hodgson, V.E. 1983. Natural learning and managerial action:
A phenomenological study in the field setting. Journal of Management
Studies, 20,3: 387-399.
Burkhardt, M. E., & Brass, D. J. 1990. Changing patterns of patterns of change:
The effects of a change in technology on social network structure and power.
Administrative Science Quarterly, 35, 104-127.
Burns, L., & Wholey, D. R. 1993. Adoption and abandonment of matrix
management programs: Effects of organizational characteristics and
interorganizational networks. Academy of Management Journal, 36,1: 106-
138.
Burt. R. 1992. Structural Holes: The Social Structure of Competition.
Cambridge: Harvard University Press.
Busch, P. & Bush, R.F. 1974. Women contrasted to men in the industrial
salesforce: Job satisfaction, values, role clarity, performance, and propensity
to leave. Journal of Marketing Research, 11: 254-260.
Cain, C. n.d. Becoming a non-drinking alcoholic: A case study in identity
acquisition. Anthropology Department. University of North Carolina,
Chapel Hill.
Campbell, J. P., McCloy, R.A., Oppler, S.H., & Sager, C.E. 1992. A theory of
performance. In N. Schmitt & W. Borman (eds.) Personnel Selection in
Organizations. San Francisco, CA: Jossey-Bass, 35-70.
REFERENCES 245
Carlile, P.R. 1997. Understanding Knowledge Transformation in Product

Development: Making Knowledge Manifest through Boundary Objects. Ph.D.
Dissertation, University of Michigan.
Carlile, P.R. 2002. A pragmatic view of knowledge and boundaries: Boundary
objects in new product development. Organization Science, 13,4: 442-455.
Carmines, E.G. & Zeller, R.A. 1979. Reliability & Validity Assessment.
Beverly Hills: Sage.
Carter, A.P. 1989. Knowhow trading as economic exchange. Research Policy,
18: 155-163.
Chidambaram, L. & Jones, B. 1993. Impact of communication medium and
computer support on group perceptions and performance: A comparison of
face-to-face and dispersed meetings. MIS Quarterly, 17,4: 465-516.
Child, J. 1972. Organizational structure, environment and performance: The
role of strategic choice. Sociology, 6: 1-22.
Choi, J.H. & Danowski, J. 2002. Making a global community on the net:
Global village or global metropolis? A network analysis of usenet
newsgroups. JCMC, 7,3.
Churchill, G.A., Jr., N.M. Ford, S.W. Hartley, & O.C. Walker, Jr. 1985. The
determinants of salesperson performance: A meta-analysis. Journal of
Marketing Research, 22: 103-118.
Cohen, W. M. & Levinthal, D. A. 1990. Absorptive capacity: A new perspective
on learning and innovation. Administrative Science Quarterly, 35: 128-152.
Coleman, J.S. 1973. The Mathematics of Collective Action. London:
Heinemann Educational Books.
Constant, D., Kiesler, S., & Sproull, L. 1994. What's mine is ours, or is it? A
study of attitudes about information sharing. Information Systems Research,
5,4: 400-422.
Constant, D., Sproull, L.S. & Kiesler, S. 1996. The kindness of strangers: The
usefulness of electronic weak ties for technical advice. Organization Science,
7: 119-135.
Contu, A. & Wilmott, H. 2000, Comment on Wenger and Yanow. Knowing in
Practice: A ‘delicate’ flower in the organizational learning field.
Cook, S.D.N. & Brown, J.S. 1999. Bridging epistemologies: The generative
dance between organizational knowledge and organizational knowing.
Organization Science, 10,4: 381-400.
Cook, S. & Yanow, D. 1993. Culture and organizational learning. Journal of
Management Inquiry, 2,4: 373-390.
Crane, D. 1972. Invisible Colleges. Chicago: The University of Chicago Press.
Cronin, B. & Rosenbaum, H. 1994. Netelligence: Informal know-how trading on
the internet. International Journal of Information Management, 14: 144-145.
Cross, R., Rice, R.E., & Parker, A. 2001b. Information seeking in social context:
Structural influences and receipt of information benefits. IEEE Transactions
246 REFERENCES
on Systems, Man, and Cybernetics-Part C: Applications and Reviews, 31, 4:

438-448.
Crossan, M.M. & Guatto, T. 1996. Organizational learning research profile.
Journal of Organizational Change Management, 9,1: 107-112.
Cudeck, R. & Browne, M.W. 1983. Cross-validation of covariance structures.
Multivariate Behavioral Research, 18: 147-167.
Czepiel, J.A. 1975. Patterns of interorganizational communications and the
diffusion of a major technological innovation in a competitive industrial
community. Academy of Management Journal, 18, 1: 6-24.
Davenport, E. 2002. Organizational knowledge and communities of practice.
Annual Review of Information Science and Technology, 171-226.
Davenport, T. H. & Prusak, L. 1998. Working Knowledge. Boston: Harvard
Business School Press.
David, F.R. & Cochran, D.S. 1987. Characteristics of boundary spanning
communicators. Journal of Technical Writing and Communication, 17,2:
165-177.
Davies, J. & Easterby-Smith, M. 1984. Learning and developing from
managerial work experiences. Journal of Management Studies, 21,2: 169-
183.
Davis, K. 1953. A method of studying communication patterns in organizations.
Personnel Psychology, 6: 301-312.
Debackere, K. & Rappa, M.A. 1994. Technological communities and the
diffusion of knowledge: A replication and validation. R&D Management,
24,4: 355-371.
Degenne, A., & Forsé, M. 1999. Introducing Social Networks. London: Sage.
Delmar, F. & Davidsson, P. (eds.) Tillväxtföretagen (High Growth Firms).
Stockholm: SNS Förlag (In Swedish).
DeMeyer, A., 1991. Tech talk: How managers are stimulating global R&D
communication. Sloan Management Review, 32,3: 49-58.
Denning, S. 2001. The Springboard: How Storytelling Ignites Action in
Knowledge-Era Organizations. Boston: Butterworth-Heinemann.
Dern, D. 1999. Usenet: Still useful after all these years. Byte, April: 5.
DeSanctis, G. & Fulk, J. 1999. Shaping Organization Form: Communication,
Connection, and Community. Thousand Oaks: Sage.
Derrida, J. 1978. Writing and Difference. London: Routledge & Kegan Paul.
Dixon, N.M. 2000. Common Knowledge: How Companies Thrive by Sharing
What they Know. Cambridge: Harvard Business School.
Donath, J., Karahalios, K & Viégas, F. 1999. Visualizing conversation. JCMC,
4,4.
Doty, D. H., Glick, W. H., & Huber, G. P. 1993. Fit, equifinality, and
organizational effectiveness: A test of two configurational theories. Academy
of Management Journal, 36: 1196-1250.
REFERENCES 247
Doz, Y. & Hamel, G. 1997. Alliance Advantage. Cambridge, MA: Harvard

Dubin, R. 1956. Industrial workers’ worlds. Social Problems, 3: 131-142.
Durkheim, E. 1893 (translated 1933, 1960). The Division of Labor in Society.
New York: Macmillan.
Earl, M. 2001. Knowledge management strategies: Toward a taxonomy.
Journal of Management Information Systems, 18,1: 215-233.
Easterby-Smith, M., Araujo, L., & Burgoyne, J. 1998. Organizational
Learning: Developments in Theory and Practice. London: Sage.
Eisenhardt, K. M. 1989a. Building theories from case study research. Academy
of Management Review, 14,4: 532-550.
Eisenhardt, K.M. 1989b. Making fast strategic decisions in high-velocity
environments. Academy of Management Journal, 32,3:543-576.
Eisenhardt, K.M. & Santos, F.M. 2002. Knowledge-based view: A new theory
of strategy? In A. Pettigrew, H. Thomas, & R. Whittington (eds.) Handbook
of Strategy and Management. London: Sage, 139-164.
Ekeh, P.P. 1974. Social Exchange Theory: The Two Traditions. London:
Heinemann Educational Books.
Etzioni, A. 1996. The responsive community: A communitarian perspective.
American Sociological Review¸ 61,1: 1-11.
Everett, M.G. & Borgatti, S.P. 1999. Peripheries of cohesive subsets. Social
Networks, 21: 397-407.
Faraj, S. & Wasko, M. M. 1998. The Web of Knowledge: An Investigation of
Self-Organizing Communities of Practice on the Net. Paper presented at the
INFORMS, Seattle, Washington.
Farr, R. 1989. The social and collective nature of representations. In J.P.
Forgas & J.M. Innes (eds.) Recent Advances in Social Psychology: An
International Perspective. Amsterdam: North-Holland Elsevier Science, 157-
166.
Feldman, M. 1987, Electronic mail and weak ties in organizations. Office:
Technology and People, 3, 3: 83-101.
Fiol, C.M. 1991. Managing culture as a competitive resource: An identity-based
view of sustainable competitive advantage. Journal of Management, 17,1:
191-211.
Flap, H. Bulder, B., & Völker, B. 1998. Intra-organizational networks and
performance: A review. Computational & Mathematical Organization
Theory, 4,2: 109-147.
Fleck, L. 1935, 1979. Genesis and Development of Scientific Fact. Chicago:
The University of Chicago Press.
Fleming, W.G. 1994. Methodography: The study of student learning as situated
action. In Gibbs, G. (ed.) Learning Approaches Evaluation and Strategy.
Improving Student Learning through Assessment and Evaluation. Oxford:
Oxford Brookes University, 525-544.
248 REFERENCES
Follett, M.P. 1924. Creative Experience. New York: Longmans, Green and
Company.
Forbes 1998. Forbes. November.
Fox, S. 1987. Self Knowledge and Personal Change: The Reported Experience
of Managers in Part-time Management Education. PhD Thesis, Manchester
University Press.
Fox, S. 2000. Communities of practice, Foucault and actor-network theory.
Journal of Management Studies, 37: 853-867.
Friedkin, N. 1980. A test of structural features of Granovetter’s strength of
weak ties theory. Social Networks, 2: 411-422.
Friedkin, N. 1982. Information flow through strong and weak ties in
interorganizational social networks. Social Networks, 3: 273-285.
Fulk, J. & DeSanctis, G. 1995. Electronic communication and changing
organizational forms. Organization Science, 6: 337-350.
Fulk, J., Flanagin, A. J., Kalman, M.E., Monge, P.R., & Ryan, T. 1996.
Connective and communal public goods in interactive communication
systems. Communication Theory, 6, 1: 60-87.
Galegher, J., Sproull, L, & Kiesler, S. 1998. Legitimacy, authority, and
community in electronic support groups. Written Communication, 15,4: 493-
530.
Galaskiewicz, J. 1996. The ‘new network analysis’ and its application to
organizational theory and behavior. In D. Iacobucci (ed.) Networks in
Marketing. Thousand Oaks: Sage, 19-31.
Galbraith, J.R. 1973. Designing Complex Organizations. Reading, MA:
Addison Wesley.
Galbraith, J. R. 1977. Organization Design. Reading, MA: Addison-Wesley.
Garfield, E., Malin, M.V., & Small, H. 1978. Citation data as science indicators.
In Y. Elkana, Y. et al. (eds.) Toward a Metric of Science: The Advent of
Science Indicators. New York: Wiley, 179-207.
Garvey, W.D. 1979. The role of scientific communication in the conduct of
research and the creation of scientific knowledge. Communication: The
Essence of Science. Oxford: Pergamon Press, 1-39.
Geertz, C. 1973. Interpretation of Cultures: Selected Essays. New York: Basic
Books.
Gergen, K.J. 1985. The social constructionist movement in modern psychology.
American Psychologist, 40,3: 266-275.
Gerstberger, P. & Allen, T.J. 1968. Criteria used by research and development
engineers in the selection of an information source. Journal of Applied
Psychology, 52: 272-279.
Gertzl, B.G. 1961. Determinants of occupational community in high status
occupations. Sociological Quarterly, 2: 37-40.
REFERENCES 249
Gherardi, I.S., Nicolini, D. & Odella, F. 1998. Toward a social understanding

of how people learn in organizations: The notion of shared curriculum.
Management Learning, 29,3: 273-297.
Gherardi, I.S. & Nicolini, D. 2000. The organizational learning of safety in
communities of practice. Journal of Management Inquiry, 9,1: 7-18.
Gherardi, I.S. & Nicolini, D. 2002. Learning the trade: A culture of safety in
practice. Organization, 9,2: 191-223.
Ghoshal, S., Korine, H., & Szulanski, G. 1994. Interunit communication in
multinational corporations. Management Science, 40,1: 96-110.
Giddens, A. 1979. Central Problems in Social Theory: Action, Structure, and
Contradiction in Social Analysis. Berkeley: University of California Press.
Gladwell, M. 2000. The Tipping Point: How Little Things Can Make a Big
Difference. Boston: Little, Brown.
Goody, E.N. (ed.). 1995. Social Intelligence and Interaction: Expressions and
Implications of the Social Bias in Human Intelligence. Cambridge:
Cambridge University Press.
Granovetter, M. S. 1973. The strength of weak ties. American Journal of
Sociology, 91: 481-510.
Granovetter, M. S. 1983. The strength of weak ties: A network theory revisited.
Sociological Theory, 1: 201-233.
Grant, R.M. 1996a. Prospering in dynamically-competitive environments:
Organizational capability as knowledge integration. Organization Science,
7,4: 375-387.
Grant, R. M. 1996b. Toward a knowledge-based theory of the firm. Strategic
Management Journal, 17 (Special Issue): 109-122.
Gupta, A.K. & Govindarajan, V. 1991. Knowledge flows and the structure of
control within multinational corporations. Academy of Management Review,
16,4: 768-792.
Haas, P. 1992. Introduction: epistemic communities and international policy
coordination. International Organization, 46,1: 1-37.
Hackman, J.R. 1976. Group influences on individuals. In M.D. Dunnette (ed.)
Handbook of Industrial and Organizational Psychology. Chicago: Rand
McNally, 1455-1518.
Hagel, J. & Armstrong, A.G. 1997. Net Gain: Expanding Markets through
Virtual Communities. Boston: Harvard Business School Press.
Hagström, P. 2000. Relaxing the boundaries of the firm. In J. Birkinshaw & P.
Hagström (eds.) The Flexible Firm. Oxford: Oxford University Press.
Hagstrom, W.O. 1965. The Scientific Community. New York: Basic Books.
Hall, R.H. 1963. The concept of bureaucracy: An empirical assessment.
American Journal of Sociology, 69: 32-40.
Han, S.K. 1996. Structuring relations in on-the-job networks. Social Networks,
18: 47-67.
250 REFERENCES
Handy, C. 1991. The Age of Unreason. Cambridge: Harvard Business School

Press.
Hansen, M. 1996. Knowledge Integration in Organizations. Ph.D. Dissertation,
Graduate School of Business, Stanford University.
Hansen, M. 1999. The search-transfer problem: The role of weak ties in sharing
knowledge across organization subunits. Administrative Science Quarterly,
44: 82-111.
Hansen, M., Nohria, N., & Tierney, T. 1999. What’s your strategy for managing
knowledge? Harvard Business Review, Jan/Feb.
Hara, N. 2000. Social Construction of Knowledge in Professional Communities
of Practice: Tales in Courtrooms. Unpublished doctoral dissertation.
Bloomington: Indiana University.
Hara, N. & Kling, R. 2002. Communities of practice with and without
information technology. ASIST 2002 Contributed Paper: 338-349.
Harary, F. 1969. Graph Theory. Reading, MA: Addison-Wesley.
Hardin, R. 1982. Collective Action. Baltimore: Johns Hopkins University Press.
Harris, M.M & Schaubroeck, J. 1988. A meta-analysis of self-supervisor, self-
peer, and peer-supervisor ratings. Personnel Psychology, 41: 43-62.
Hauptman, O. 1986. The influence of task type on the relation between
communication and performance: The case of software development. R & D
Management, 16, 2: 127-139.
Hedlund, G. 1994. A model of knowledge management and the N-form
corporation. Strategic Management Journal, 15.
Hedlund, G. & Nonaka, I. 1993. Models of knowledge management in the West
and Japan. In Lorange, P., Chakravarthy, B., Roos, J. & Van de Ven, A.,
(eds.) Implementing Strategic Process, 117-145. Blackwell Business.
Heneman, H.G. III. 1974. Comparisons of self- and superior ratings of
managerial performance. Journal of Applied Psychology, 59,5: 638-642.
Hildreth, P. & Kimble, C. (eds.). 2004 (forthcoming). Knowledge Networks:
Innovation Through Communities of Practice. London: Idea Group Inc.
Hildreth, P., Kimble, C., & Wright, P. 2000. Communities of practice in the
distributed international environment. Journal of Knowledge Management,
4,1: 27-38.
Hill, S. 1981. Competition and Control at Work: The New Industrial Sociology.
London: Heinemann Educational Books.
Hinds, P. & Kiesler, S. 1995. Communication across boundaries: Work,
structure, and use of communication technologies in a large organization.
Organization Science, 6: 373-393.
Holzner, B. & Marx, J.H. 1979. Knowledge Application: The Knowledge
System in Society. Boston: Allyn & Bacon, Inc.
Homans, G.C. 1950. The Human Group. New York: Harcourt Brace.
Huber, G. P. 1991. Organizational learning: The contributing processes and
literatures. Organization Science, 2,1: 88-115.
REFERENCES 251
Huber, G. P., Miller, C. C., & Glick, W. H. 1990. Developing more

encompassing theories about organizations: The centralization-effectiveness
relationship as an example. Organization Science, 1,1: 11-40.
Hutchins, E. 1991. Organizing work by adaptation. Organization Science, 2,1:
14-39.
Hutchins, E. 1995. Cognition in the Wild. Cambridge: MIT Press.
Hutchins, E. 1996. Learning to navigate. In S. Chaiklin & J. Lave (eds.)
Understanding Practice: Perspectives on Activity and Context. New York:
Jablin, F.M. 1979. Superior-subordinate communication: The state of the art.
Psychological Bulletin, 86,6: 1201-1222.
Jablin, F. M. 1987. Organizational communication theory and research: An
overview of communication climate and network research. In D. Nimmo
(ed.), Communication Yearbook 4. New Brunswick, NJ: Transaction Books.
Jarvenpaa, S.L. & Staples, D.S. 2000. The use of collaborative electronic media
for information sharing: An exploratory study of determinants. Strategic
Information Systems, 9: 129-154.
Jarvenpaa, S.L. & Staples, D.S. 2001. Exploring perceptions of organizational
ownership of information and expertise. Journal of Management Information
Systems, 18,1: 151-183.
Johnson, C.M. 2001. A survey of current research on online communities of
practice. Internet and Higher Education, 4: 45-60.
Johnson, J. D. 1993. Organizational Communication Structure. Norwood, NJ:
Ablex.
Jordan, B. 1989. Cosmopolitical obstetrics: Some insights from the training of
traditional midwives. Social Science and Medicine, 28, 9: 925-944.
Judge, W.Q., Fryxell, G.E., & Dooley, R.S. 1997. The new task of R&D
management: Creating goal-directed communities for innovation. California
Management Review, 39,3: 72-85.
Kaplan, S., Schenkel, A., von Krogh, G., & Weber, C. 2002. Knowledge-based
theories of the firm in strategic management: A review and extension. In A.
Schenkel, Communities of Practice or Communities of Discipline: Managing
Deviations at the Öresund Bridge. Published Ph.D. Thesis, Stockholm
School of Economics.
Katz, D. & Kahn, R. 1966. The Social Psychology of Organizations. New
York: John Wiley & Sons, Inc.
Kerlinger, F.N. 1986. Foundations of Behavioral Research. Orlando: Harcourt
Brace.
Kettinger, W.J. & Grover, V. 1997. The use of computer-mediated
communication in an interorganizational context. Decision Sciences, 28, 3:
513-555.
252 REFERENCES
Kiesler, S. & Cummings, J. 2002. What do we know about proximity in work

groups? A legacy of research on physical distance. In P. Hinds & S. Kiesler
(eds.) Distributed Work. Cambridge: MIT Press.
Kiesler, S. & Sproull, L. 1986. Response effects in the electronic survey.
Public Opinion Quarterly, 50: 402-413.
Kim, A.J. 2000. Community Building on the Web: Secret Strategies for
Successful Online Communities. Berkeley: Peachpit Press.
Klein, K.J. & D’Aunno, T.A. 1986. Psychological sense of community in the
workplace. Journal of Community Psychology, 14: 365-377.
Klein, J. K. Dansereau, F., & Hall, R.J. 1994. Level issues in theory
development, data collection, and analysis. Academy of Management Review,
19,2: 195-229.
Kling, R. 1980. Social analyses of computing: Theoretical perspectives in
recent empirical research. Computing Surveys, 12, 1: 61-110.
Knorr-Cetina, K.D. 1981. The Manufacture of Knowledge: An Essay on the
Constructivist and Contextual Nature of Science. Oxford: Pergamon Press.
Kochen, M. (ed.) 1989. The Small World. Norwood, NJ: Ablex.
Kogut, B. 2000. The networks as knowledge: Generative rules and the
emergence of structure. Strategic Management Journal, 21: 405-425.
Kogut, B. & Zander, U. 1992. Knowledge of the firm, combinative capabilities
and the replication of technology. Organization Science, 3,3: 383-397.
Kogut, B. & Zander, U. 1996. What firms do? Coordination, identity and
learning. Organization Science, 7,5: 502-518.
Kollock, P. 1998. Social dilemmas: The anatomy of cooperation. Annual Review
of Sociology, 24: 183-214.
Kollock, P. 1999. The economies of online cooperation: Gifts, and public goods
in cyberspace. In M. A. Smith & P. Kollock (eds.), Communities in
Cyberspace. London: Routledge, 220-239.
Kollock, P & Smith, M. 1996. Managing the virtual commons: Cooperation
and conflict in computer communities. In S.C. Herring (ed.), Computer-
mediated Communication: Linguistic, Social and Cross-Cultural
Perspectives. Amsterdam: John Benjamins, 109-128.
Komorita, S.S. & Parks, C.D. 1995. Interpersonal relations: Mixed motive
interaction. Annual Review of Psychology, 46: 183-207.
Krackhardt, D. 1990. Assessing the political landscape: Structure, cognition, and
power in organizations. Administrative Science Quarterly, 35: 342-369.
Krackhardt, D. 1994. Constraints on the interactive organization as an ideal
type. In C. Heckscher & A. Donnellon (eds.), The Post-Bureaucratic
Organization: New Perspectives on Organizational Change. Thousand Oaks,
CA: Sage, 211-222.
Krackhardt, D., Blythe, J., & McGrath, C. 1994. Krackplot 3.0: An improved
network drawing program. Connections, 17, 2: 53-55.
REFERENCES 253
Krackhardt, D. & Brass, D. 1994. Intraorganizational networks: The micro

side. In S. Wasserman & J. Galaskiewicz (eds.) Advances in Social Network
Analysis. London: Sage, 207-229.
Krackhardt, D., & Hanson, J. R. 1993. Informal networks: The company behind
the chart. Harvard Business Review, 71: 104.
Krackhardt, D., & Porter, L. W. 1985. When friends leave: A structural analysis
of the relationship between turnover & stayers' attitudes. Administrative
Science Quarterly, 30: 242-261.
Kraut, R., Patterson, M., Lundmark, V., Kiesler, S., Mukopadhyay, T., &
Scherlis, W. 1998. Internet paradox: A social technology that reduces social
involvement and psychological well-being? American Psychologist, 53,9:
1017-1031.
Kreiner, K. & Schultz, M. 1993. Informal collaboration in R&D. The
formation of networks across organizations. Organization Studies, 14,2: 189-
209.
Krikorian, D.H., Seibold, D.R., & Goode, P.L. 1997. Reengineering at LAC: A
case study of emergent network processes. In B.D. Sypher (ed.), Case Studies
in Organizational Communication 2: Perspectives on Contemporary Work
Life. New York: Guilford Press, 129-144.
Kuhn, T. 1962. The Structure of Scientific Revolution. Chicago: University of
Chicago Press
Kunda, G. 1992. Engineering Culture: Control and Commitment in a High-
tech Organization. Philadelphia: Temple University Press.
Lakhani, K. & von Hippel, E. 2000. How open source software works: Free
user-to-user assistance. Paper presented at The 3rd Intangibles Conference.
Knowledge: Management, Measurement and Organization, Stern School of
Business, NYU.
Latour, B. 1987. Science in Action: How to Follow Scientists and Engineers
through Society. Cambridge: Harvard University Press..
Lave, J. 1988. Cognition in Practice: Mind, Mathematics and Culture in
Everyday Life. Cambridge: Cambridge University Press.
Lave, J. & Wenger, E. 1991. Situated Learning: Legitimate Peripheral
Participation. Cambridge: Cambridge University Press.
Lawrence, T.B. 1998. Examining resources in an occupational community:
Reputation in Canadian forensic accounting. Human Relations, 51, 9: 1103-
1131.
Lazega, E. 2001. The Collegial Phenomenon: The Social Mechanisms of
Cooperation among Peers in a Corporate Law Partnership. Oxford: Oxford
University Press.
Lee, A.S. 1994. Electronic mail as a medium for rich communication: An
empirical investigation using hermeneutic interpretation. MIS Quarterly, 18,
2: 143-157.
254 REFERENCES
Leifer, R. & Huber, G.P. 1977. Relations among perceived environmental

uncertainty, organizational structure, and boundary-spanning behavior.
Leonard-Barton, D. 1992. Core capabilities and core rigidities: A paradox in
managing new product development. Strategic Management Journal, 13
(Summer Special Issue): 111-126.
Lesser, E.L., Fontaine, M.A. & Slusher, J.A. (eds.). 2000. Knowledge and
Communities. Boston: Butterworth-Heinemann.
Lesser, E. & Storck, J. 2001. Communities of practice and organizational
performance. IBM System Journal, 40: 831-841.
Levitt, B. & March, J. G. 1988. Organizational learning. Annual Review of
Sociology, 14: 319-340.
Levy, D.A. & Nail, P.R. 1993. Contagion: A theoretical and empirical review
and reconceptualization. Genetic, Social, and General Psychology
Monographs, 119: 235-284.
Liebeskind, J.P., Oliver, A.L., Zucker, L., Brewer, M. 1996. Social networks,
learning, and flexibility: Sourcing scientific knowledge in new biotechnology
firms. Organization Science, 7,4: 429-443.
Liedtka, J. 1999. Linking competitive advantage with communities of practice.
Journal of Management Inquiry, 8: 5-16.
Liefeld, J. P. 1988. Response effects in computer-administered questioning.
Journal of Marketing Research, 25, 4: 405-409.
Lievrouw, L.A., Rogers, E.M., Lowe, U, & Nadel, E. 1987. Triangulation as a
research strategy for identifying invisible colleges among biomedical
scientists. Social Networks, 9: 217-248.
Lin, N., Ensel, W. M. & Vaughn, J.C.. 1981. Social resources and strength of
ties: Structural factors in occupational status attainment. American
Sociological Review, 46: 393-405.
Lissoni, F. 2001. Knowledge codification and the geography of innovation: The
case of Brescia mechanical cluster. Research Policy, 30: 1479-1500.
Lueg, C. & Fisher, D. 2003. From Usenet to CoWebs: Interacting with Social
Information Spaces. London: Springer.
Macdonald, S. 1995. Learning to change: An information perspective on
learning in the organization. Organization Science, 6,5: 557-568.
Macdonald, S. & Williams, C. 1993. Beyond the boundary: An information
perspective on the role of the gatekeeper in the organization. Journal of
Production Innovation Management, 10: 417-427.
Mansfield, E. 1985. How rapidly does new industrial technology leak out?
Journal of Industrial Economics, 34,2: 217-223.
Marsick, V.J. & Watkins, K.E. 1990. Informal and Incidental Learning in the
Workplace. London: Routledge.
REFERENCES 255
Martinez, J.I. & Jarillo, J.C. 1989. The evolution of research on coordination
mechanisms in multinational corporations. Journal of International Business
Studies, Fall: 489-514.
Marsden, P. 1990. Network data and measurement. In W.R. Scott & J. Blake
(eds.). Annual Review of Sociology, 16: 435-463.
Marsden, P. 1998. Memetics and social contagion: Two sides of the same coin?
Journal of Memetics – Evolutionary Models of Information Transmission, 2,
68-85.
Marshall, H. 1972. Structural constraints on learning. In B. Geer (ed.) Learning
to Work. Beverly Hills, CA: Sage Publications.
Maynard, Jr. H.B. & Mehrtens, S.E. 1993. The Fourth Wave: Business in the
21rst Century. San Francisco, CA: Berrett-Koehler.
Maznevski, M.L. & Chudoba, K.M. 2000. Bridging space over time: Global
virtual team dynamics and effectiveness. Organization Science, 11,5: 473-
492.
McDermott, R. 1999a. Learning across teams: How to build communities of
practice in team organizations. Knowledge Management Review, 8: 32-38.
McDermott, R. 1999b. Why information technology inspired but cannot deliver
knowledge management. California Management Review, 41,3: 103-117.
Medoff, P. & Sklar, H. 1994. Streets of Hope: The Fall and Rise of an Urban
Neighborhood. Boston, MA: South End Press.
Mehra, A., Kilduff, M., & Brass, D.J. 2001. The social networks of high and
low self-monitors: Implications for workplace performance. Administrative
Miles, R. E., & Snow, C.C. 1986. Organizations: New concepts for new forms.
California Management Review, 28,3: 62-73.
Miles, R. E., & Snow, C.C. 1992. Causes of failure in network organizations.
California Management Review, Summer: 53-72.
Miles, R. E., & Snow, C.C. 1995. The new network firm: A spherical structure
built on a human investment philosophy. Organizational Dynamics, 23: 5-18.
Miller, C. C., Glick, W. H., Wang, Y. D., & Huber, G. P. 1991. Understanding
technology-structure relationships: Theory development and meta-analytic
theory testing. Academy of Management Journal, 34,2: 370-399.
Mintzberg, H. 1973. The Nature of Managerial Work. New York: Harper Row.
Monge, P. R. 1995. Global network organizations. In R. Cesaria & P. Shockley-
Zalabak (eds.), Organization Means Communication. Rome, Italy: SIPI, 135-
151.
Monge, P. & Contractor, N. 1997. Emergence of communication networks. In
Jablin, F. M., & Putnam, L.L. (eds.) Handbook of Organizational
Communication (2nd ed.). Thousand Oaks, CA: Sage.
Monge, P. & Contractor, N. 2003. Theories of Communication Networks.
Oxford University Press.
256 REFERENCES
Monge, P. R., & Eisenberg, E. M. 1987. Emergent communication networks. In

F. M. Jablin, L.L. Putnam, K.H. Roberts, and L.W. Porter, (eds.), Handbook
of Organizational Communication. Newbury Park, CA: Sage, 304-342.
Monge, P. R., Fulk, J., Kalman, M.E., Flanagin, A.J., Parnassa, C., & Rumsey,
S. 1998. Production of collective action in alliance-based interorganizational
communication and information systems. Organization Science, 9,3: 411-433.
Monge, P. Rothman, L., Eisenberg, E., Miller, K., & Kirste, K. 1985. The
dynamics of organizational proximity, Management Science, 31,9: 1129-
1141.
Musgrave, R.A. 1959. The Theory of Public Finance. New York: McGraw-Hill.
Nadler, D & Tushman, M. 1988. Strategic Organization Design: Concepts,
Tools, & Processes. Glenview: Scott, Foresman.
Nahapiet, J. & Ghoshal, S. 1998. Social capital, intellectual capital, and the
organizational advantage. Academy of Management Review, 23,2: 242-266.
Newsted, P.R. 1985. Paper versus online presentations of subjective
questionnaires. International Journal of Man-Machine Studies, 23, 231-247.
Nicolini, D. & Meznar, M.B. 1995. The social construction of organizational
learning: Conceptual and practical issues in the field. Human Relations, 48,7:
727-746.
Nonaka, I. 1994. A dynamic theory of organizational knowledge creation.
Organization Science, 5, 1: 14-37.
Nonnecke, B. & Preece, J. 2003. Silent participants: Getting to know lurkers
better. In C. Lueg & D. Fisher (eds.) From Usenet to CoWebs: Interacting
with Social Information Spaces. London: Springer.
Nonnecke, B. & Preece, J. 2000. Lurker demographics: Counting the silent.
Proceedings of CHI’2000. Hague, The Netherlands: 73-80.
Nunnally, J.C. & Bernstein, I.H. 1978. Psychometric Theory. New York:
McGraw-Hill.
Oliver, A. & Liebeskind, J.P. 1998. Three levels of networking for sourcing
intellectual capital in biotechnology. Int. Studies of Management &
Oliver, P. E., Marwell, G. & Teixeira, R. 1985. A theory of critical mass I:
Group heterogeneity, interdependence and the production of collective goods.
American Journal of Sociology, 91: 522-556.
Olson, M. 1965. The Logic of Collective Action. Cambridge, MA: Harvard
University Press.
Orlikowski, W. J. 1996. Learning from Notes: Organizational issues in
groupware implementation. In R. Kling (ed.), Computerization and
Controversy. New York: Academic Press, 173-189.
Orlikowski, W.J. 2002. Knowing in practice: Enacting a collective capability in
distributed organizing. Organization Science, 13,3: 249-273.
REFERENCES 257
Orlikowski, W.J., Ang, S., Weill, P., Krcmar, H.C. & DeGross, J.I. (eds.). 2000.
The Proceedings of the 21st International Conference on Information
Systems, Brisbane, Australia, ICIS, 313-328.
Orr, J. 1990. Talking about Machines: An Ethnography of a Modern Job. Ph.D.
Thesis, Cornell University.
Orr, J. 1996. Talking about Machines: An Ethnography of a Modern Job.
Cornell University.
Ortner, S.B. 1989. High Religion: A Cultural and Political History of Sherpa
Buddhism. Princeton: Princeton University Press.
Palloff, R.M. & Pratt, K. 1999. Building Learning Communities in Cyberspace:
Effective Strategies for the Online Classroom. San Francisco: Jossey-Bass.
Palmer, D., Friedland, R., & Singh, J. V. 1986. The ties that bind:
Organizational and class bases of stability in a corporate interlock network.
American Sociological Review, 51,6: 781-796.
Parsons, T. 1951. The Social System. Glencoe, IL: Free Press.
Peck, M.S. 1987. The Different Drum. New York: Simon & Schuster.
Pelz, D.C. & Andrews, F.M. 1966. Scientists in Organizations. New York: John
Wiley & Co.
Perreault, W.D. Jr. & L.E. Leigh. 1989. Reliability of nominal data based on
qualitative judgments. Journal of Marketing Research, 26: 135-148.
Pickering, J. M. & King, J. L. 1995. Hardwiring weak ties: Interorganizational
computer-mediated communication, occupational communities, and
organizational change. Organization Science, 6: 479-486.
Pinch, T.J. 1990. The role of scientific communities in the development of
science. Impact of Science on Society, 159: 219-225.
Pinch, T.J. & Bijker, W.E. 1984. The social construction of facts and artefacts:
Or how the sociology of science and the sociology of technology might
benefit each other. Social Studies of Science, 14: 399-441.
Polanyi, M. 1962a. Personal Knowledge: Towards a Post-Critical Philosophy.
New York:
Polanyi, M. 1962b. The republic of science. Minerva, 1:54-73.
Price, D.J. de S. 1963. Little Science, Big Science. New York: Columbia
University Press.
Price, D.J. de S. 1965. Is technology historically independent of science? A
study in statistical historiography. Technology and Culture, 6: 553-568.
Pugh, D.S., Hickson, D.J., Hinings, C.R., Turner, C. 1968. Dimensions of
organization structure. Administrative Science Quarterly, 13: 65-105.
Pugh, D.S., Hickson, D.J., Hinings, C.R., & Turner, C. 1969. The context of
organization structures. Administrative Science Quarterly, 14: 91-114.
Quan-Haase, A. 2002. Information Brokering, Performance, and Culture: A
Case Study of an Internet-era Firm. Thesis proposal. Toronto: Faculty of
Information Studies, University of Toronto.
258 REFERENCES
Quan-Haase, A., & Wellman, B. (in press). How does the Internet affect social
capital? In M.H. Huysman & V. Wulf (eds.) Information Technology and
Social Capital. Boston: Kluwer Academic Publishers.
Ranson, S., Hinings B., & Greenwood, R. 1980. The structuring of
organizational structures. Administrative Science Quarterly, 25: 1-17.
Reber, A.S. 1993. Implicit Learning and Tacit Knowledge, New York: Oxford
University Press.
Rennecker, J. 2001. The situated nature of distributed collaborative work.
Sloan School of Management, MIT, Cambridge.
Resnick, L.B., Levine, J.M., & Teasley, S.D. (eds.). 1991. Perspectives on
Socially Shared Cognition. Washington, DC: American Psychological
Association.
Rich, G.A., Bommer, W.H., MacKenzie, S.B., Podsakoff, P.M., & Johnson, J.L.
1999. Apples and apples or apples and oranges? A meta-analysis of
objective and subjective measures of salesperson performance. Journal of
Personal Selling & Sales Management, 19,4: 41-52.
Ridings, C.M, Gefen, D., Arinze, B. 2002. Some antecedents and effects of trust
in virtual communities. Strategic Information Systems, 11: 271-295.
Roberts, K.H., & O'Reilly, C.A. 1978. Organizations as communication
structures: An empirical approach. Human Communication Research, 4,4:
283-293.
Roethlisberger, F., & Dickson, W. 1939. Management and the Worker. New
York: Wiley.
Rogers, E.M. 1995. Diffusion of Innovations. New York, NY: The Free Press.
Sack, W. 2000. Conversation map: A content-based usenet newsgroup
browser. ACM International Conference on Intelligent User Interfaces –
2000 IUI’00: 233-240.
Salaman, G. 1974. Community and Occupation: An Exploration of
Work/Leisure Relationships. London: Cambridge University Press.
Sampson, E.E. 1981. Cognitive psychology as ideology. American
Psychologist, 36,7: 730-743.
Samuelson, P. A. 1954. The pure theory of public expenditure. Review of
Economics and Statistics, 36,4: 387-389.
Sarason, S.B. 1974. The Psychological Sense of Community: Prospects for a
Community Psychology. San Francisco: Jossey-Bass.
Saxenian, A. 1996. Regional Advantage: Culture, competition in Silicon Valley
and Route 128, Second edn. Cambridge, MA: Harvard University Press.
Schank, R.C. 1995. Tell Me a Story: Narratives and Intelligence. Evanston, IL:
Northwestern University Press.
Schenkel, A. 2002. Communities of Practice or Communities of Discipline:
Managing Deviations at the Öresund Bridge. Published Ph.D. dissertation,
Stockholm School of Economics.
REFERENCES 259
Schenkel, A. & Teigland, R. 2002. Investigating the relationship between

communities of practice and organizational performance. In A. Schenkel,
Communities of Practice or Communities of Discipline: Managing
Deviations at the Öresund Bridge. Published Ph.D. dissertation, Stockholm
School of Economics.
Schenkel, A., Teigland, R. & Borgatti, S.P. 2001. Theorizing structural
properties of communities of practice: A social network approach. Presented
at Academy of Management Conference.
Schott, T. 1988. International influence in science: Beyond center and
periphery. Social Science Research, 17: 219-238.
Schrader, S. 1991. Informal technology transfer between firms: Cooperation
through information trading. Research Policy, 20: 153-170.
Schutz, A. 1970. In H.R. Wagner (ed.) Alfred Schutz on Phenomenology and
Social Relations. Chicago: University of Chicago Press.
Schwartz, S. H. 1970. Moral decision making and behavior. In J. Macauley &
L. Berkowitz (eds.) Altruism and Helping Behavior: Social Psychological
Studies of Some Antecedents and Consequences. New York: Academic Press,
127-141.
Scott, R.K. 1995. Creative employees: A challenge to managers. Journal of
Creative Behavior, 29,1: 64-71.
Scott, W.R. 1998. Organizations: Rational, Natural, and Open Systems, Fourth
edn. Upper Saddle River: Prentice-Hall, Inc.
Shmanske, S. 1991. Public Goods, Mixed Goods, and Monopolistic
Competition. College Station: Texas A&M University Press.
Singer, J. 2003. Small world networks in R&D: A study of the factors which
promote diverse knowledge sharing ties among scientists. Working Paper.
Smith, M. 1999. Invisible crowds in cyberspace: Measure and mapping the
social structure of USENET. In Smith, M. & Kollock, P. (eds.) Communities
in Cyberspace. London: Routledge.
Smith, M. & Kollock, P. 1999. Communities in Cyberspace. London:
Routledge.
Snyder, W. 1996. Organizational Learning and Performance: An Exploration
of Linkages Between Organizational Learning, Knowledge, and Performance.
University of Southern California. Ph.D. Dissertation.
Sole, D. & Edmondson, A. 2002. Situated knowledge and learning in dispersed
teams. British Journal of Management, 13: S17-S34.
Spender, J.-C. 1996. Making knowledge the basis of a dynamic theory of the
firm. Strategic Management Journal, 17: 45-62.
Sproull, L. & Faraj, S. 1995. Atheism, sex and databases: The net as a social
technology. In B. K. J. Keller (ed.), Public Access to the Internet. Cambridge,
MA: MIT Press, 62-81.
Sproull, L. & Kiesler, S. 1986. Reducing social context cues: Electronic mail in
organizational communication. Management Science, 32,11: 1492-1512.
260 REFERENCES
Sproull, L. S. & Kiesler, S. B. 1991. Connections: New Ways of Working in the

Networked Organization. Cambridge, MA: MIT Press.
Squire, K.D. & Johnson, C.B. 2000. Supporting distributed communities of
practice with interactive television. Educational Technology Research and
Development, 48,1:23-43.
Star, S.L. & Griesemer, J.R. 1989. Institutional ecology, ’translations’ and
boundary objects: Amateurs and professionals in Berkeley’s Museum of
Vertebrate Zoology, 1907-1939. Social Studies of Science, 19: 387-420.
Steinfield, C. 1992. Computer-mediated communications in organizational
settings. Management Communication Quarterly, 5,3: 348-365.
Stevenson, W.B. & Gilly, M.C. 1993. Problem-solving networks in
organizations: Intentional design and emergent structure. Social Science
Research, 22: 92-113.
Storck, J. & Hill, P.H. 2000, Knowledge diffusion through ‘Strategic
Communities’. Sloan Management Review, 39,4:81-93.
Strauss, A. 1978. A social world perspective. Studies in Symbolic Interaction.
1: 119-128.
Suchman, L. 1983. Office procedure as practical action: Models of work and
system design. ACM Transactions on Office Information Systems, 1,4: 320-
328.
Swan, J., Scarbrough, H. & Robertson, M. 2002. The construction of
‘communities of practice’ in the management of innovation. Management
Learning, 33,4: 477-496.
Szulanski, G. 1996. Exploring internal stickiness: Impediments to the transfer of
best practice within the firm. Strategic Management Journal, 17: 27-43.
Taylor, R.L. 1975. The technological gatekeeper. R & D Management, 5: 239-
242.
Taylor, R.L. and Utterback, J. 1975. A longitudinal study of communication in
research. IEEE Transactions on Engineering Management, EM-22,2: 80-87.
Teece, D.J. 1998. Capturing value from the knowledge assets: The new
economy, markets for know-how, and intangible assets. California
Management Review, 40, 3: 55-79.
Teigland, R. 2000. Communities of practice at an internet firm: Netovation vs.
on-time performance. In E. L. Lesser, M.A. Fontaine, & J.A. Slusher (eds.),
Knowledge and Communities. Boston: Butterworth-Heinemann: 151-178.
Teigland, R. 2002. Exploring the relationships between network of practice
participation, centrality, and individual performance in a multinational
organization. Presented at INSNA – International Network of Social Network
Analysts Sunbelt Conference.
Teigland, R., Fey, C. F. & Birkinshaw, J.M. 2000. Knowledge dissemination in
global R&D operations: An empirical study of multinationals in the high-
technology industry. Management International Review, Special issue on
International Management of Technology, 1.
REFERENCES 261
Teigland, R. & Wasko, M.M. 2004 (forthcoming). Creative ties and ties that
bind: The impact of participation in networks of practice on individual
performance. In P.M. Hildreth (ed.). Knowledge Networks: Innovation
Through Communities of Practice. London: Idea Group Inc.
Teigland, R. & Wasko, M.M. 2003 (forthcoming). Integrating knowledge
through information trading: Examining the impact of boundary spanning
communication on individual performance. Decision Sciences, Special Issue
on Knowledge Management.
Thibaut, J. W., & Kelley, H. H. 1959. The Social Psychology of Groups. New
York: John Wiley.
Thompson, J.D. 1967. Organizations in Action: Social Science Bases of
Administrative Theory. New York: McGraw-Hill.
Thorn, B. K. & Connolly, T. 1987. Discretionary data bases. A theory and
some experimental findings. Communication Research, 14, 5: 512-528.
Tönnies, F. 1887, translated 1957. Gemeinschaft und Gesellschaft (Community
and Society). East Lansing, MI: Michigan State University Press.
Tsai, W. 2002. Social structure of “Coopetition” within a multiunit
organization: Coordination, competition, and intraorganizational knowledge
sharing. Organization Science, 13,2: 179-190.
Tsai, W. & Ghoshal, S. 1998. Social capital and value creation: The role of
intrafirm networks. Academy of Management Journal, 41,4: 464-476.
Tsoukas, H. 1996. The firm as a distributed knowledge system: A
constructionist approach. Strategic Management Journal, 17, Winter Special
Issue: 11-25.
Tsoukas, H. & Vladimirou, E. 2001. What is organizational knowledge?
Journal of Management Studies, 38,7: 973-993.
Tuire, P. & Erno, L. 2001. Exploring invisible scientific communities: Studying
networking relations within an educational research community. A Finnish
case. Higher Education, 42: 493-513.
Tuomi, I. 1999. Corporate Knowledge: Theory and Practice of Intelligent
Organizations. Helsinki: Metaxis.
Turban, D.B. & Dougherty, T.W. 1994. Role of protégé personality in receipt of
mentoring and career success. Academy of Management Journal, 37, 3: 688-
702.
Tushman, M. 1977. Special boundary roles in the innovation process.
Tushman, M.L. & Katz, R. 1980. External communication and project
performance: An investigation into the role of gatekeepers. Management
Science, 26,11:1071-1085.
Tushman, M.L. & Scanlan, T.J. 1981. Characteristics and external orientations
of boundary spanning individuals. Academy of Management Journal,
24,1:83-98.
262 REFERENCES
Tyler, J.R., Wilkinson, D.M. & Huberman, B.A. 2003. Email as spectroscopy:
Automated discovery of community structure within organizations. Working
paper, HP Labs.
Usdiken, B. & Pasadeos, Y. 1995. Organizational analysis in North America
and Europe: A comparison of co-citation networks. Organization Studies,
6,3: 503-526.
Van Maanen, J. & Barley, S. R. 1984. Occupational communities: Culture and
control in organizations. In B. M. Staw & L. L. Cummings (eds.) Research in
Organizational Behavior, Vol. 6. Greenwich, CO: JAI Press.
von Hippel, E. 1987. Cooperation between rivals: Informal know-how trading.
Research Policy, 16: 291-302.
von Hippel, E. 1988. The Sources of Innovation. New York: Oxford University
Press.
von Hippel, E. 2001. Innovation by user communities: Learning from open
source software. Sloan Management Review, Summer: 82-86.
von Hippel, E. & Schrader, S. 1996. Managed informal information trading:
The oil scout system in oil exploration firms. International Journal of
Technology Management, 11: 207-218.
von Hippel, E. & von Krogh, G. 2003. Open Source Software and the Private-
Collective Innovation Model: Issues for Organization Science. Organization
Science, forthcoming.
von Krogh, G. 1998. Care in knowledge creation. California Management
Review¸ 40,3: 133-153.
von Krogh, G. 2002. The communal resource and information systems. Journal
of Strategic Information Systems, 11: 85-107.
von Krogh, G., Roos, J., & Slocum, K. 1994. An essay on corporate
epistemology. Strategic Management Journal, 15: 53-71.
von Krogh, G. & Roos, J. 1995. Organizational Epistemology. New York: St.
Martin’s Press.
Vygotsky, L.S. 1962. Thought and Language. Cambridge: MIT Press.
Walker, M., Wasserman, S., & Wellman, B. 1994. Statistical models for social
support networks. In Wasserman, S., & Galaskiewicz, J. (eds.), Advances in
Social Network Analysis. London: Sage.
Wasko, M. 2002. Why Should I Share? Examining Knowledge Contribution in
Networks of Practice. Unpublished doctoral dissertation. Graduate School of
the University of Maryland at College Park.
Wasko, M. & Faraj, S. 1999. Webs of knowledge: An examination of
knowledge types and knowledge flows in an electronic community of
practice. Paper presented at the Academy of Management Conference.
Wasko, M. & Faraj, S. 2000. It is what one does: Why people participate and
help others in electronic communities of practice. Journal of Strategic
Information Systems, 9, 2-3: 155-173.
REFERENCES 263
Wasko, M. & Faraj, S. 2003. Why should I share? Examining knowledge

contribution in networks of practice. Working paper, College of Business,
Florida State University.
Wasko, M. & Teigland, R. 2002. The provision of online public goods:
Examining social structure in a network of practice. In The Proceedings of
the 23rd International Conference on Information Systems, Barcelona, Spain,
ICIS.
Wasserman, S. & Faust, K. 1994. Social Network Analysis: Methods and
Applications. Cambridge: Cambridge University Press.
Weber, M. 1946. Bureaucracy. In H.H. Gerth & C.W. Mills (eds.) From Max
Weber: Essays in Sociology. New York: Oxford University Press, 196-244.
Weber, M. 1947. Max Weber: The Theory of Social and Economic
Organization. New York: Oxford University Press. Translated by A.M.
Henderson and T. Parsons.
Weber, M. 1978 (German edition 1922). Economy and Society (Wirtschaft und
Gesellschaft). Berkeley: University of California.
Weedman, J. 1993. On the ‘isolation’ of humanists: A report of an invisible
college. Communication Research, 20,6: 749-776.
Weekley, J.A. & Gier, J.A. 1989. Ceilings in the reliability and validity of
performance ratings: The case of expert raters. Academy of Management
Journal, 32, 1: 213-222.
Weick, K. 1979. The Social Psychology of Organizing. Menlo Park: Addison-
Wesley.
Welbourne, T.M., Johnson, D.E., & Erez, A. 1998. The role-based
performance scale: Validity analysis of a theory-based measure. Academy of
Management Journal, 41,5: 540-555.
Wellman, B. 2001. Physical place and cyberplace: The rise of personalized
networking. International Journal of Urban and Regional Research, 25,2:
227-252.
Wellman, B. & Berkowitz, S.D. 1988. Introduction: Studying Social Structures.
In B. Wellman & S.D. Berkowitz (eds.), Social Structures: A Network
Approach. Volume 15 in Contemporary Studies in Sociology. New York:
Wellman, B., Salaff, J., Dimitrova, D., Garton, L., Gulia, M., &
Haythornthwaite, C. 1996. Computer networks as social networks:
Collaborative work, telework, and virtual community. Annual Review of
Sociology, 22: 213-238.
Wellman, B. & Gulia, M. 1999. Net surfers don't ride alone. In P. Kollock & P.
Smith (eds.), Communities in Cyberspace. Berkeley: University of California
Press.
Wellman, B., Quan-Haase, A., Witte, J., & Hampton, K. 2001. Does the Internet
increase, decrease, or supplement social capital? Social networks,
264 REFERENCES
participation, and community commitment. American Behavioral Scientist,

45, 3: 437-456.
Wenger, E. 1998. Communities of Practice: Learning, Meaning, and Identity,
Cambridge: Cambridge University Press.
Wenger, E. 2000a. Communities of practice: The key to knowledge strategy. In
E.L. Lesser, M.A. Fontaine, & J.A. Slusher (eds.), Knowledge and
Communities. Oxford: Butterworth-Heinemann, 3-20.
Wenger, E. 2000b. Communities of practice and social learning systems.
Organization, 7,2:225-246.
Wenger, E. & Snyder, W.M. 2000. Communities of Practice: The
Organizational Frontier. Harvard Business Review, January-February: 139-
145.
Wenger, E. McDermott, R, & Snyder, W.M. 2002. Cultivating Communities of
Practice. Boston: Harvard Business School Press.
Wexley, K.N., Alexander, R.A., Greenawalt, J.P. & Couch, M.A. 1980.
Attitudinal congruence and similarity as related to interpersonal evaluations in
manager-subordinate dyads. Academy of Management Journal, 23,2: 320-
330.
Williams, R.L, & Cothrel, J. 2000. Four smart ways to run online communities.
Sloan Management Review, Summer: 81-91.
Williamson, O.E. 1991. Comparative economic organization: The analysis of
discrete structural alternatives. Administrative Science Quarterly, 36,2: 269-
296.
Wittgenstein, L. 1953. Philosophical Investigations. Oxford: Basil Blackwell.
Yanow, D. 2000. Seeing organization learning: A cultural view. Organization,
7,2: 247-268.
Yin, R.K. 1989. Case Study Research: Design and Methods. London: Sage
Publications.
Zander, U. & Kogut, B. 1995. Knowledge and the speed of the transfer and
imitation of organizational capabilities: An empirical test. Organization
Science, 6,1: 76-92.
Ziman, J. 1978. Reliable Knowledge: An Exploration of the Grounds for Belief in
Science. Cambridge: Cambridge University Press.
APPENDIX ONE
A Look at Knowledge
THIS SECTION PRESENTS some of the current perspectives relating to

knowledge, e.g., dimensions of knowledge (individual vs. collective, tacit vs.
explicit) and knowing. This discussion is not meant to be exhaustive of the
perspectives relating to knowledge, rather the idea is to give the reader an initial
understanding of these perspectives as related to the purpose of this thesis.
1.1 Dimensions of Knowledge
Since its appearance in the 1970s in connection with the organizational learning
literature, research on “knowledge” has continued to grow. In recent literature,
researchers have tended to focus their discussions on two dimensions of
knowledge: (1) the tacit / explicit dimension and (2) the individual / collective or
group dimension (Cook & Brown, 1999). We discuss each of these dimensions
separately before looking at the relationships between these.
The tacit-explicit dimension was one of the first distinctions of knowledge to be

made in the organizational literature. This distinction has its roots in the work
by the philosopher Ryle (1949:25), who distinguishes between “knowing that”
of theory and “knowing how” of practice. Polanyi (1962:56) further expanded
on this to discuss the two inseparable aspects of knowledge: “knowing what”
and “knowing how”. Based on Polanyi’s work, Nelson & Winter (1982) then
introduced this tacit-explicit distinction into the organizational literature when
they defined tacit knowledge as “knowledge that cannot be articulated” (ibid:
76). In essence, explicit knowledge is that which has been codified and refers to
knowledge that is transmittable in formal, systematic language. Tacit
knowledge, on the other hand, has a personal quality and is deeply embedded in
action and context, making it difficult to formalize and communicate. Other
distinctions along these lines include that by Kogut & Zander (1992) who
discuss “information” and “know-how”. Information is defined as “knowledge
that can be transmitted without loss of integrity once the syntactical rules
266 APPENDIX ONE
required for deciphering it are known” while know-how is based upon von
Hippel’s (1988: 76) definition: “know-how is the accumulated practical skill or
expertise that allows one to do something smoothly and efficiently”. Polanyi’s
(1962:49) example of bike-riding provides an example of this tacit-explicit
distinction. Individuals who know how to ride a bike know which way to turn
the handlebars when they are riding in order to keep their balance. However,
they have difficulty in explaining to another individual which way they turn the
handlebars to remain upright. Thus, what Polanyi called the tacit dimension is
what the individual knows in terms of maintaining balance while the explicit
dimension is what the individual can communicate to another about riding a
bike.
Whether tacit and explicit knowledge are two distinct forms of knowledge or the
ends of a continuum is one area of debate among researchers. In his framing of
“knowing how” and “knowing that”, Ryle (1949) described these as two distinct
kinds of knowledge. This view is adapted by Brown & Duguid (1998) since
they propose that “know-how” is different from “know-what” in dispositional
character. Cook & Brown (1999:384) also argue that, “explicit and tacit
knowledge are two distinct forms of knowledge (i.e., neither is a variant of the
other); that each does work the other cannot; and that one form cannot be made
out of or changed into the other.” However, Taylor (1993) proposes that tacit
and explicit knowledge exist on a kind of a continuum. Tacit knowledge
provides the background understanding on which an individual’s explicit
knowledge rests.
In recent years, the second dimension – individual/collective - has gained

increasing attention. The traditional Cartesian view considers knowledge as
held by the individual since it sees the individual thinker as the primary wielder
and repository of what is known (Cook & Brown, 1999). This view is presented
in Polanyi’s (1962) work since he viewed knowledge at the individual level and
originating in individual intuition, as even reflected in the title of his book,
Personal Knowledge. This individual focus is also represented in Simon’s
(1991:125) statement that “All learning takes place inside individual human
heads”.
However, of late the volume of research and publication focusing on a more

sociological approach to knowledge has grown considerably. This approach
sees knowledge as socially constructed and embedded in the social relationships
between individuals (Kogut & Zander, 1992) and as distributed across groups or
collectives of individuals who develop and possess social relationships.
Durkheim (1893) serves as the ascribed source for the origins of knowledge as a
A LOOK AT KNOWLEDGE 267
social construction, arguing in his work that individuals are not only
independent psychological decision-making entities, but they are also purely
social beings within a “collective conscience” (1960: 283). Another source of
the sociological perspective is work by Fleck (1935, 1979) on thought
collectives. In his work first published in 1935, Fleck provided an insightful
description of knowledge communities, combining the creativity and
socialization aspects of thought. Fleck’s main thesis was that the development
of knowledge is a social phenomenon and that knowing, thinking, and
knowledge creation are not something that an individual is capable of doing
alone. Instead, these activities take place in thought collectives – sociological
groups with a common style of thinking (Haas, 1992; Tuomi, 1999)60.
In line with this, Weick & Roberts (1994) discussed a ‘collective mind’ in the
organizational literature while examples of an interest in the collective level
include work such as that by Wenger (1998) and Brown & Duguid (1991, 1998)
in which they discuss communities of practice, by Kogut & Zander (1992, 1996)
in their view of the firm as a social community, and by Nonaka & Takeuchi
(1995) in their spiral of organizational knowledge creation. Spender (1996) also
discusses this dimension, building on Durkheim (1893). In addition, von Krogh
and co-authors (1994, 1995) have investigated epistemological issues at the
collective level, discussing such terms as “organizational knowledge” and
“organizational epistemology”. The main argument of this perspective on
knowledge is that a collective or group of individuals possesses a “body of
knowledge” that the individual member of the collective does not61. For
example, a group of copier repair technicians possesses a body of knowledge
about copy repair that is held in common by the various technicians; however,
each individual technician does not possess the entire body of knowledge (Cook
& Brown, 1999).
60
Fleck’s original work was widely neglected since it proposed an all too unconventional way
of looking at science. Instead of science being based on objective facts independent of any
social processes, Fleck proposed that scientific facts only make sense within a given style of
thought that is learned through socialization into the worldview of a specific thought
community (Tuomi, 1999). Fleck’s work was further developed by Knorr-Cetina (1981) and
Pinch & Bijker (1984) into the social constructivist view of science. This view holds that
science, as the product of human activity, is not objective, but rather that this knowledge is
“constructed” by researchers whose perceptions of reality are shaped by their training, beliefs,
and life experiences (Lievrouw et al., 1987).
61
There are numerous definitions of knowledge using a sociological perspective: (1) socially
spread and influenced by social settings (Schutz, 1970), (2) a social construction, embedded
in a system of individual, lasting relationships (Berger & Luckman, 1966; Weick, 1979), (3)
based on the interaction of several meanings (Derrida, 1978), (4) shared by “agents who
process data” through cultural processes (Boisot, 1995), (5) material, but also mental and
social (Latour, 1987). See also Blackler (1993, 1995) for a further discussion.
268 APPENDIX ONE
The two dimensions of tacit/explicit and individual/collective have been

incorporated into models by several researchers. Spender (1996) developed a
two-by-two matrix, creating four types of knowledge: conscious (explicit
knowledge held by the individual), objectified (explicit knowledge held by the
collective), automatic (preconscious individual knowledge), and collective
(highly context-dependent knowledge that is manifested in the collective’s
activities). In this manner, he argued that the psychological individual type of
tacit knowledge that was examined by Polanyi could be separated from the
sociological or collective type. An extension of this 2x2 model to the firm has
been made by several scholars. Hedlund & Nonaka (1993) and Kogut & Zander
(1992) present models of organizational knowledge in which information and
know-how are on one axis and individual, group, organization, and network on
the other.
Several models have also been proposed in order to account for the dynamics of
complex organizational work. In their model of organizational knowledge
creation, Nonaka (1994) and Nonaka & Takeuchi (1995) discuss the interactions
between the two dimensions, arguing that a firm’s knowledge creation is
performed by individuals and cannot occur without these individuals. However,
the organization plays a critical role in articulating and amplifying this
knowledge. In this model, knowledge creation spirals between tacit and explicit
knowledge and moves between the individual, group, organization, and inter-
organizational levels. One of their key arguments is that knowledge creation
occurs when individuals share and develop knowledge through social interaction
with others throughout the formal and informal levels of the organization
(Nonaka, 1994).
1.2 Knowing
Cook & Brown (1999) further argue that a static discussion of the four types of
knowledge merely provides us with an understanding of the knowledge
possessed by people. To say that “ ‘Robert knows auto mechanics” points to
Robert possessing knowledge of auto mechanics” (p.382, italics in original).
Yet in order to give a full account of what individuals know, we need to focus
on both the knowledge they possess as well as the actions they perform using the
knowledge they have. Merely describing the knowledge that individuals or
groups have does not provide us with an understanding of how this knowledge is
used when individuals or groups take action in their work. Thus, researchers are
now widening their focus of knowledge to include knowing, a verb connoting
action, doing, and practice as well as knowledge, a noun connoting things,
elements, facts, processes, etc. (Orlikowski, 2002). Cook & Brown (1999:387,
italics in original) defined the concept of “knowing” as “not something that is
used in action or something necessary to action, but rather something that is a
part of action (both individual and group)”. Thus, knowing is part of the actual
work performed such as when an auto mechanic tunes an engine.
While there is a growing interest in knowing, researchers have differing

opinions regarding this concept. Building on the American Pragmatist
perspective and in particular the work of the philosopher John Dewey (1934),
Cook & Brown see knowing as being distinct from knowledge and thus distinct
from the tacit form of knowledge. However, Orlikowski proposes that tacit
knowledge is a form of knowing, inseparable from action since it is constituted
through such action, basing her argument on the work of Ryle (1949), Polanyi
(1966), and Schön (1983). Whether tacit knowledge and knowing are distinct
will continue to be an area for debate, but what is more essential to the
introduction of knowing into the organizational literature is the argument that
knowing (and tacit knowledge depending on one’s definition) and action are
mutually constituted. Empirical studies of work practices have provided
evidence of this recursive relationship (Suchman, 1987; Lave, 1988; Orr, 1996).
Knowing is thus argued to be an ongoing social accomplishment that is
constituted and reconstituted in everyday work practice (Orlikowski, 2002).
In order to emphasize the relationship between knowledge and/or knowing and

action at work, researchers have developed the practice-based perspective. The
main idea of the practice-based perspective is that it emphasizes the collective,
situated, and provisional nature of knowledge in contrast to the rational-
cognitive view of knowledge (Sole & Edmondson, 2002). The object of focus is
practice as a means to capture the nature of knowledge in action in
organizations, and researchers have developed concepts such as “knowledge-in-
practice” (Carlile, 1997), “knowing in practice” (Orlikowski, 2002), and
“epistemology of practice” (Cook & Brown, 1999). The notion of practice has a
long intellectual history. John Dewey, mentioned above, argued that knowledge
is manifested most rigorously not in what you claim to understand, but what you
can do (see Schön, 1983). In addition, in social theory Bourdieu (1977)
developed the nature of practice in his seminal work, Outline of a Theory of
Practice. Practice implies the actions of individuals and groups when
conducting real work, e.g., the practice of software engineers, nurses, hotel
managers, etc. Practice is an activity, an interaction among individuals (Lave,
1988) doing their “real work” as it is informed by a particular organizational or
270 APPENDIX ONE
group context (Cook & Brown, 1999:387)62. Cook & Brown (1999) distinguish
practice from behavior and action, explaining that behavior is “doing of any
sort” and action is behavior instilled with meaning, while practice refers to
action informed by meaning within a particular group context. The differences
are best explained by example provided by Cook & Brown (1999). If my knee
jerks, then this is behavior, and if I tap my knee with a physician’s hammer to
check my reflexes, then this is action. However, if a physician taps my knee
with a hammer as part of my yearly check-up, then this is practice.
Researchers argue that if we are to understand the means with which

organizations generate knowledge as well as new ways of using knowledge, we
need to focus on both knowledge as well as practice and their interplay (Cook &
Brown, 1999)63. When individuals and groups interact with the world through
performing their practice, the four forms of knowledge are brought into play.
Through this interaction between action and knowledge, new knowledge as well
as new ways of using knowledge are then generated. As a group of individuals
collaborate on a common task, they apply a common body of knowledge.
Through applying this common knowledge, they also increase their knowledge
since knowledge and learning are difficult to separate in dynamic task
environments. An individual learns by participating in an activity and thus
builds his or her knowledge. Thus, social interactions and not the isolated
behavior of individuals are the basis for knowledge creation, exchange,
evaluation, and integration. As such, knowledge creation and learning are
situated, social processes that cannot be separated from working with knowledge
as the outcome (Brown & Duguid, 1991; Lave & Wenger, 1991; Purser,
Pasmore, & Tenkasi, 1992; Boland & Tenkasi, 1995; Wenger, 1998).
62
For a review of distributed practice at the organizational level, see Tsoukas (1996).
63
As mentioned earlier, for discussions and reviews of organizational learning, see Huber
(1991), Crossan & Guatto (1996), and Easterby-Smith et al. (1998).
REFERENCES
Boland, R. J., Jr. & Tenkasi, R. V. 1995. Perspective making and perspective
taking in communities of knowing. Organization Science, 6, 4: 350-372.
Bourdieu, P. 1977. Outline of a Theory of Practice. Cambridge: Cambridge
University Press.
Review, 40,3: 90-111.
Carlile, P.R. 1997. Understanding Knowledge Transformation in Product
Development: Making Knowledge Manifest through Boundary Objects. Ph.D.
Dissertation, University of Michigan.
Cook, S.D.N. & Brown, J.S. 1999. Bridging epistemologies: The generative
dance between organizational knowledge and organizational knowing.
Dewey, J. 1934. Art as Experience. New York: Minton, Balch.
Durkheim, E. 1893 (translated 1933, 1960). The Division of Labor in Society.
New York: Macmillan.
Fleck, L. 1935, 1979. Genesis and Development of Scientific Fact. Chicago:
The University of Chicago Press.
Haas, P. 1992. Introduction: epistemic communities and international policy
coordination. International Organization, 46,1: 1-37.
Hedlund, G. & Nonaka, I. 1993. Models of knowledge management in the West
and Japan. In Lorange, P., Chakravarthy, B., Roos, J. & Van de Ven, A.,
(eds.) Implementing Strategic Process, 117-145. Blackwell Business.
Lave, J. 1988. Cognition in Practice: Mind, Mathematics and Culture in
Everyday Life. Cambridge: Cambridge University Press.
Monge, P. & Contractor, N. 1997. Emergence of communication networks. In
Jablin, F. M., & Putnam, L.L. (eds.) Handbook of Organizational
Communication (2nd ed.). Thousand Oaks, CA: Sage.
Nanda, A. 1996. Resources, capabilities and competencies. In B. Moingeon &
A. Edmondson (eds.) Organizational Learning and Competitive Advantage.
Thousand Oaks, CA: Sage, 93-120.
Nelson, R.R. & Winter, S.G. 1982. An Evolutionary Theory of Economic
Change. Cambridge: Belknap Press.
272 APPENDIX ONE

Nonaka, I. & Takeuchi, H. 1995. The Knowledge Creating Company, New
York: Oxford University Press.
Orlikowski, W.J. 2002. Knowing in practice: Enacting a collective capability in
distributed organizing. Organization Science, 13,3: 249-273.
Orr, J. 1990. Talking about Machines: An Ethnography of a Modern Job. Ph.D.
Thesis, Cornell University.
Cornell University.
Polanyi, M. 1962. Personal Knowledge: Towards a Post-Critical Philosophy.
New York:
Polanyi, M. 1966. The Tacit Dimension. New York: Doubleday.
Purser, R.E., Pasmore, W.A. & Tenkasi, R.V. 1992. The influence of
deliberations on learning in new product development teams. Journal of
Engineering and Technology Management, 9: 1-28.
Ryle, G. 1949. The Concept of Mind. London: Hutchinson.
Schön, D. 1983. The Reflective Practitioner. New York: Basic Books.
Simon, H. A. 1991. Bounded rationality and organizational learning.
Sole, D. & Edmondson, A. 2002. Situated knowledge and learning in dispersed
teams. British Journal of Management, 13: S17-S34.
Suchman, L. 1987. Plans and Situated Actions: The Problem of Human
Machine Communication. Cambridge: University of Cambridge Press.
Taylor, C. 1993. To follow a rule… In Calhoun, C., LiPuma, E. & Postone, M.
(eds.) Bourdieu: Critical Perspectives. Chicago: University of Chicago Press.
York: John Wiley.
Thorn, B. K. & Connolly, T. 1987. Discretionary databases. A theory and some
experimental findings. Communication Research, 14, 5: 512-528.
Tuomi, I. 1999. Corporate Knowledge: Theory and Practice of Intelligent
Organizations. Helsinki: Metaxis.
von Hippel, E. 1988. The Sources of Innovation. New York: Oxford University
Press.
von Krogh, G. 1998. Care in knowledge creation. California Management
Review¸ 40,3: 133-153.
von Krogh, G. & J. Roos. 1995. Corporate Epistemology. New York: St.
Martin’s Press.
von Krogh, G., Roos, J., & Slocum, K. 1994. An essay on corporate
epistemology. Strategic Management Journal, 15: 53-71.
Wasko, M. 2002. Why Should I Share? Examining Knowledge Contribution in

Networks of Practice. Unpublished doctoral dissertation. Graduate School of
the University of Maryland at College Park.
Weick, K. & Roberts, K.H. 1993. Collective mind in organizations: Heedful
interrelating on flight decks. Administrative Science Quarterly, 38,3: 357-
381.
APPENDIX TWO
Empirical Studies
Article 1
Theorizing Structural Properties of Communities of Practice: A Social

Network Approach
Article 2
The Provision of Online Public Goods: Examining Social Structure in a

Network of Practice

Conference on Information Systems (ICIS), Barcelona, Spain, 2002.
Article 3
Communities of Practice in a High-Growth Internet Consultancy:

Netovation vs. On-Time Performance
By R. Teigland

276 APPENDIX TWO


In F. Delmar & P. Davidsson (eds.), Tillväxtföretagen (High-Growth Firms),

Stockholm: SNS Förlag, 2001 (in Swedish).
Article 4
Extending Richness with Reach: Participation and Knowledge Exchange in

Electronic Networks of Practice
By R. Teigland & M.M. Wasko
To be published in P. Hildreth & C. Kimble (eds.),Knowledge Networks:

Innovation Through Communities of Practice, London: Idea Group Inc., 2004
(forthcoming).
Previous version published in W.J. Orlikowski, S. Ang, P. Weill, H.C. Krcmar,

& J.I. DeGross (eds.), The Proceedings of the 21st International Conference on
Information Systems, Brisbane, Australia, ICIS, 2000.
Article 5
Integrating Knowledge Through Information Trading: Examining the

Impact of Boundary Spanning Communication on Individual Performance

APPENDIX TWO 277
Article 6
Exploring the Relationships Between Network of Practice Participation,

Centrality, and Individual Performance in a Multinational Organization
By R. Teigland
Previous version presented at INSNA – International Network of Social Network

Analysts Sunbelt Conference, 2002.
Article 7
Knowledge Dissemination in Global R&D Operations: An Empirical Study

of Multinationals in the High-Technology Industry
By R. Teigland, C.F. Fey, & J.M. Birkinshaw
Published in Management International Review, Special Issue on International

Management of Technology, 2000, 1.
Note to Reader:
As we have progressed through the research presented in these studies, we have

used a variety of different terms for more or less the same group of ideas. If we
were writing a monograph instead of a compendium of articles, we could have
easily resolved this problem by ensuring that the terminology is consistent by
going back and making the necessary corrections throughout the entire thesis.
However, as the majority of these articles are published, this task is not possible.
Thus, in order to try and prevent any confusion for the reader, we feel that it is
necessary to provide an overview of the different terms that are used throughout
the empirical studies. At the beginning of each individual article, we clarify the
terms that we have used in the article and how they relate to the terms that we
use in the thesis.
Article One
Theorizing Structural Properties of Communities of

Practice: A Social Network Approach
Note to Reader on Terminology:
In general, the terminology in this article corresponds to the terminology in this

thesis. We use one abbreviation:
1) CP for Community of practice

280 ARTICLE ONE
Previous version presented at Academy of Management, 2001
Theorizing Structural Properties of Communities of

Practice: A Social Network Approach
Andrew Schenkel
Stockholm School of Economics, Box 6501,113 83 Stockholm, Sweden
email: andrew.schenkel@hhs.se
Robin Teigland
Stockholm School of Economics, Box 6501,113 83 Stockholm, Sweden
email: robin.teigland@hhs.se, URL: www.teigland.com
Stephen P. Borgatti
Carroll School of Management, Boston College, Chestnut Hill, MA 02167 USA
email: borgatts@bc.edu
ABSTRACT
Management researchers are paying considerable attention to communities of practice

(CPs) as a means to better understand the creation, transfer, and embedding of
knowledge in organizations. However, scant attention has been paid to understanding
the structural dimensions of these organizational forms. Thus, this study’s purpose is
to conceptualize the structural properties of communities of practice. We draw on the
social network literature and apply some of the extensively used network concepts and
measures to develop five structural properties for CPs. We illustrate the usefulness of
these properties through applying them to data from a complex construction project.
Finally, we develop a series of propositions that link the structural properties of CPs to
organizational performance before concluding with a discussion.
Keywords: community of practice, social network, knowledge, performance,

construction
THEORIZING STRUCTURAL PROPERTIES 281
INTRODUCTION
Due to considerable changes in the competitive environment during recent decades, an
increasing number of scholars are suggesting that knowledge is perhaps the only “true”
source of competitive advantage for a firm (Drucker, 1991; Kogut & Zander, 1992;
Spender & Grant, 1996). Many of these scholars have chosen to anchor their work in
the knowledge-based view of the firm. Within this perspective, the concept of
communities of practice (CPs), or emergent, informal groups that form through the
mutual engagement in a shared practice (Brown & Duguid, 1991; Lave & Wenger,
1991), is rapidly gaining attention. The concept of CPs is related to several key
knowledge activities and offers considerable promise for several reasons. First, CP
research provides insight into how the informal organizational structures are the nexus
for the sharing and transfer of valuable individual and group tacit knowledge (Kogut &
Zander, 1992). Second, this informal organizational structure also provides a
protective capability that helps impede the transfer of valuable knowledge to outside
the firm (Kogut & Zander, 1992, 1996; Liebeskind, 1996). Third, communities of
practice provide firms with a vital source of incremental innovation as community
members continuously create knowledge to improve the practice. Finally,
communities of practice have strategic implications since researchers have noted that
the patterns of informal organizations directly affect organizational outcomes (Kotter,
1982, 1985, Kanter, 1983, 1989).
However, research on communities of practice is still in its earliest stages of
development. The majority of the CP literature has concentrated on defining
communities of practice, primarily focusing on the cognitive processes of how
communities emerge and operate and often relying on anecdotal accounts as the basis
for recent theory development (Wenger, 1998). However, researchers have paid little
attention to conceptualizing the structural properties of communities of practice. In
addition, few researchers have looked at how CPs impact organizational performance
and competitive advantage (Liedtka, 1999; Storck & Hill, 2000). These gaps in the
research seem surprising because on the one hand, there is such a strong relationship
between cognition and structure, and on the other hand, the informal structure has been
shown to play an important role in organizational outcomes (Kotter, 1982, 1985;
Kanter, 1983, 1989; Miles & Snow, 1994).
Thus, the purpose of this paper is twofold: 1) to conceptually develop the
structural properties of communities of practice and 2) to propose a series of
relationships between CP structural properties and performance. Through identifying
and specifying structural properties of communities of practice, we may open the door
for additional theorizing on both the structural as well as the cognitive aspects of CPs
as well as for further empirical studies. For example, we may use these structural
properties to help detect and analyze communities of practice within organizations, to
track their development over time, or to measure their relationship to organizational
performance.
282 ARTICLE ONE
To achieve the above, we draw on the extensive field of social networks since
we believe that the logic of communities of practice carries with it strong parallels
with the structural characteristics of embedded networks. The vast stream of social
network literature offers analytical tools that describe and analyze organizational
structures. We select extensively used concepts, measures, and techniques from social
network analysis based on their ability to help describe CPs and then synthesize these
with existing concepts from CP literature to develop five structural properties of
communities of practice. Finally, we propose a series of relationships between these
structural properties and performance.
This paper is organized as follows. We begin by reviewing the literature on
communities of practice and thereafter, we develop five structural properties of
communities of practice. Following that, we use a case study of a multi-billion dollar
construction project to illustrate these properties. We then develop a series of
propositions relating the structural properties to performance before concluding with a
discussion of the research implications and study limitations.
THEORETICAL BACKGROUND
In an ethnographic study of Xerox service technicians during the late 1980s, it was
observed that there was a variance between the organization’s formal description of
work and the way in which the actual work was performed (Orr, 1990). When the
technicians were faced with problems for which the formal structure often did not
provide solutions, they relied on the organization’s informal systems for help, such as
story-telling, conversation, mentoring, and experiential learning (Orr, 1990; Brown &
Duguid, 1991; Lave & Wenger, 1991; Wenger, 1997). These emergent structures have
been coined communities of practice and have been defined in the following manner –
a group of people informally and contextually bound in a work situation who are
applying a common competence in the pursuit of a common enterprise (Brown &
Duguid, 1991; Lave & Wenger, 1991; Wenger, 1998).
The work situation can be seen as the context for the process of negotiating a
common enterprise (Wenger, 1998), the pillar of any community of practice. During
this negotiation process, members engage in three cognitive processes: narration,
collaboration, and social construction (Brown & Duguid, 1991). Through the
narration of stories, employees help each other to make sense of ambiguous, problem-
centered situations and in this context noncanonical practice is exercised. Problems
are diagnosed through the building of a coherent account of a random sequence of
events, while at the same time a causal cognitive map is developed (Brown & Duguid,
1991). The second aspect of CPs is the collaboration that occurs among its members.
With knowledge-intensive tasks, often no one individual can solve the problem on his
or her own due to an individual’s bounded rationality. By relying on the community,
individuals can perform their work without needing to know everything (Wenger,
1998). The third process, social construction, occurs through the mutual engagement
of the members of a community of practice. During a process of comprehension,
members negotiate meanings, turning incoherent events into coherent accounts and
creating insights for the benefit of the community (Brown & Duguid, 1991; Wenger,
1998). In this process, the members develop a shared repertoire consisting of both the
tacit and explicit means of communication and working that enable the community to
perform its practice. In particular, the explicit means include the community’s own
language and vocabulary, codified procedures, documents, regulations, etc. But more
interestingly, the tacit means, such as the implicit relations, cues, unarticulated
etiquette, etc., are the invisible glue that holds the community together (Brown &
Duguid, 1991; Boland & Tenkasi, 1995; Wenger, 1998). In addition, through these
three cognitive processes, individuals satisfy their social needs of companionship,
belonging, identity, and status. Members become bounded together by the context of
the situation in an informal manner creating the social fabric of the organization
(Brown & Duguid, 1991). It is precisely this invisible glue that bonds the community
together and differentiates a community of practice from any other type of community
(Wenger, 1998).
In addition to the above cognitive aspects of CPs, researchers have also begun
to look at various structural aspects of CPs, and in particular, participation levels
within communities. And as far as we have been able to discern, participation levels
are the only structural aspects of communities of practice that have been discussed in
any significant detail. Thus, we include here a somewhat lengthy discussion of this
dimension since it provides a platform for some of the structural properties that we
develop later in the paper.
Wenger (1998) has suggested the following categories of community
participation: 1) full participation (insider), 2) legitimate peripherality, 3) marginality,
and 4) full non-participation (outsider). In full participation, the person is an inclusive
member of the community. He or she has gained legitimacy through engaging with
other actors of the community in common actions and has acquired the formal and
informal ability to behave as a community member (Lave & Wenger, 1991). The
member is proficient in the tacit and explicit means of communication and working
that enable the community to perform its practice. However, legitimate peripherality,
the second category, connotes a level of only partial participation in the community.
Gaining access to the periphery is not unproblematic since boundaries and entrance
requirements may exist. For example, full participants may develop close
relationships that exclude outsiders, or a complex, detailed understanding of the
community’s practice may be required to become a full participant. Thus, legitimate
peripheral participation indicates that the individual has gained some legitimacy
among full participants. An apprentice is one example of a legitimate peripheral
participant, gaining community knowledge and acceptance, and on his or her way to
becoming a full participant.
As in the case of legitimate peripherality, marginality is a mixture of
participation and non-participation. While the boundary between these two levels is
284 ARTICLE ONE
unclear, the key difference between them is the participant's trajectory in the
community (Wenger, 1998). In the case of legitimate peripherality, the person is
either on an inbound trajectory to becoming a full participant or on a circular trajectory
around the periphery. However, in the case of marginality, the person's trajectory is
outbound, and he or she is thus either moving from being a full participant to
becoming an outsider or is restricted to the periphery by the community with little
hope of becoming a full participant. Marginal participants may be best understood by
looking at practices of discrimination. In such cases, while participants wish to
become community insiders, they are continuously pushed back into identities of non-
participation (Wenger, 1998).
Finally, the opposite of full participation is full non-participation, or total
exclusion from the community. This form of participation may either be decided by
the community or by the non-participant since there is no desire to be part of the
community.
Of primary importance is that an individual's participatory status can be
considered to be a public good and not wholly owned by either party. The individual
and the community jointly and continuously determine the individual's status with
regard to participation. In addition, these levels of participation are not absolutes.
Rather, they are contextual and temporal, which means they are fluid and contingent
upon the current community configuration.
To summarize, our primary objective in this section was to discuss the two
main elements of CPs: cognitive and structural. The cognitive elements consist of
narration, social construction, and collaboration, while the structural elements connote
the different participation levels. We now turn to the field of social networks to help
conceptually develop structural properties of CPs.
DEVELOPMENT OF CP STRUCTURAL PROPERTIES

A social network has been defined as a “specific set of linkages among a defined set of
persons" (Mitchell, 1969:2). The fundamental principle in social networks is that pair-
wise relationships among individuals link to form networks whose structural
characteristics (discussed in greater detail below) are both the result of dynamic
processes and affect group and individual outcomes. At the individual level, the
notion is that a person's position in the network provides both constraints and
opportunities for the individual. At the network level, there is the holistic notion of
emergent properties that suggests that at least some properties and outcomes of a
social network are a function of its complete structure and are not reducible to either
an individual actor or a single link (Degenne & Forsé, 1993). For a review of social
network concepts and principles, see Wasserman & Faust (1994).
With regard to this connection between communities of practice and social
networks, Wenger has noted their relationship in several places (1998: 74, 126, 287,
298). In fact, he states that CPs can be viewed as nodes of “strong” ties within
interpersonal networks. However, he takes pains to point out that there is a clear
distinction between the two: "A community of practice is not defined merely by who
knows whom or who talks with whom in a network of interpersonal relations….What

is of interest to me is not so much the nature of interpersonal relationships through
which information flows as the nature of what is shared and learned and becomes a
source of cohesion – that is the structure and content of practice" (ibid: 74, 21). Thus,
what distinguishes a community of practice from other networks is that a community
of practice is a contextually based network consisting of individuals who are involved
in a common enterprise. Through this common enterprise and continuous mutual
engagement, members develop a shared repertoire of meanings and practices. Thus,
every community of practice consists of a network, but not every network forms a
community of practice.
If we agree that a community of practice is a network but not every network is a
community of practice, then the question arises whether there are specific structural
properties that are likely to distinguish a community of practice from other networks.
In this section, we develop five distinguishing structural properties of CPs: 1)
connectedness, 2) graph-theoretic distance, 3) density, 4) core/periphery structure, and
5) coreness.
Connectedness
Perhaps the most fundamental aspect of communities of practice is mutual engagement
(Wenger, 1998). Through engagement, individuals participate in each of the three
cognitive CP processes: narration, collaboration, and social construction. The result of
this interaction is a complex network of social relations and interdependency. Thus,
the extent to which members of a group are connected via pair-wise interaction ties is
an index of the extent to which the group can potentially function as a community of
practice. Individuals who are not interacting with others in a group cannot learn the
community’s practice and thus will not be identified as being members of the
community. Therefore, a minimum structural characteristic of a CP is that every
member has appropriate ties (e.g., advice-giving, trust, etc.) with some if not all other
members of the community. In other words, all community members are directly or
indirectly connected with each other and there are no isolates. In social network
analysis, the maximal set of individuals who are directly or indirectly connected to
each other in a network is called a connected component (Harary, 1969). Therefore, a
CP is necessarily located wholly within a single connected component. This then
leads us to our first structural property:
STRUCTURAL PROPERTY 1: Connectedness - In a community of practice, every

member is connected, directly or indirectly, to every other member. That is, a
community of practice is contained within a connected component.
Graph-theoretic Distance
Another fundamental characteristic of communities of practice is the notion of shared
repertoire (Wenger, 1998). Social network research has studied the diffusion (sharing)
of ideas and attitudes extensively (Friedkin, 1982; Burt, 1992; Rogers 1995). A
286 ARTICLE ONE
central tenet of this research is the notion that in both diffusion and influence
processes, the graph-theoretic distance between nodes in a network dictates the extent
to which they are expected to share ideas. The graph-theoretic distance between two
nodes is defined as the number of links in the shortest path connecting them. Thus, the
greater the graph-theoretic distances between pairs of group members, the longer it
takes for information to flow from one to the other, and the greater the likelihood that
what is transmitted arrives too late, too distorted, or fails to arrive at all. Individuals
separated by wide distances tend to develop variations (e.g., in language, values,
norms, etc.) that are not shared, contrary to the notion of a single community of
practice. Thus, communities of practice can be expected to have shorter distances on
average than organizational networks in general. Thus, we suggest the second
structural property:
STRUCTURAL PROPERTY 2: Graph-theoretic distance - Relative to organizational

networks in general, communities of practice have shorter graph-theoretic
distances between all pairs of members.
Density
Through mutual engagement and the associated cognitive processes, the practice of the
community is disseminated and developed. Connectedness is a necessary prerequisite
for this development but not sufficient in and of itself since a certain level of density is
required. The density of a network measures the degree of cohesion in the group
(Blau, 1977) and is defined as the total number of ties divided by the total number of
possible ties in the network. A dense network consists of people who are for the most
part directly connected to each other, rather than connected through intermediaries.
Direct connections are far more powerful in terms of influence and transmitting tacit
knowledge. Hence, through a dense network, a community's practice is more evenly
disseminated. In social network terms, density is a function of the average number of
contacts that each individual possesses, and it is the average number of ties per person
divided by N-1, where N is the number of individuals in the network. A community of
practice should exhibit a higher density than the organizational network in which it is
embedded, which is discussed further in structural property 4. Therefore, we propose
the following:
STRUCTURAL PROPERTY 3: Density - Relative to organizational networks in

general, communities of practice have a greater density of ties.
Core/periphery Structure
Community of practice theory distinguishes between communities and constellations
(Wenger, 1998). A constellation is a set of different communities of practice (possibly
involving overlapping membership) that have different shared repertoires and different
joint goals. Groups that have largely different membership, interact primarily within-
group rather than with members of other groups, and develop separate sets of shared
repertoire can be seen as forming a single constellation, but not a single community of
practice. Structurally then, it is obvious that communities of practice do not contain
significant subgroupings since such subgroupings will constitute separate, although
interlinked, communities of practice. This in turn implies that communities have a
core/periphery structure as is described in social network theory. A network has a
core/periphery structure to the extent that it contains no significant subgroups,
factions, or cliques except the core itself (Borgatti & Everett, 1999; Everett & Borgatti,
1999). Stated in another way, a network has a core/periphery structure if it “can be
partitioned into two sets: a core whose members are densely tied to each other, and a
periphery whose members have more ties to core members than to each other” (Everett
& Borgatti, 1999: 397.) Network researchers have developed statistical procedures for
measuring the extent to which an observed network conforms to a core/periphery
structure (Borgatti & Everett, 1999).
Core/periphery structures facilitate the diffusion of information and innovation
because they do not contain significant clusters of nodes that are poorly connected to
the rest of the network. Consequently, they can be expected to lead to a relatively
homogeneous group culture (a shared repertoire) in which most individuals are
exposed to new practices and ideas soon after they emerge. In contrast, networks that
are divided into cliques or factions work against the establishment of a single
community of practice. Different subgroups tend to develop their own norms, beliefs,
and practices, which then effectively create separate CPs that are loosely connected to
each other – i.e., constellations. Thus, we have our fourth structural property:
STRUCTURAL PROPERTY 4: Core/periphery - Communities of practice have

core/periphery structures rather than clique structures.
Coreness
As described above, Wenger (1998) distinguished full participation in a community of
practice from legitimate peripheral participation and marginal participation, and the
distinction between the latter two depends on the legitimacy of the individual. This
distinction is fundamentally cognitive rather than structural. Thus, in a network
analysis of a set of relations at a single moment, it would be difficult to distinguish
between legitimate peripheral participation and marginal participation. However, the
difference between these and full participation can be detected by the coreness
measures that are produced as a byproduct of fitting the core/periphery model
(Borgatti & Everett, 1999). Technically, coreness is defined as the principal
eigenvector of the network matrix (Bonacich, 1972). In non-mathematical terms,
coreness indicates the extent to which a node is located in the center or periphery of a
group. Nodes with high coreness are well connected to both core and peripheral
members. Nodes with low coreness are connected mostly to core members. Thus, this
structural property mimics the position of new apprentices in a community, who
initially are connected through a few experienced members who show them the ropes,
288 ARTICLE ONE
and who gradually form ties with more and more people. Hence, coreness is the basis
for our last structural property:
STRUCTURAL PROPERTY 5: Coreness - The greater an individual's participation in

a community of practice, the greater his or her coreness score.
To the extent that the above translation from social network theory to
community of practice theory is faithful, we now have a set of properties for detecting
and evaluating communities of practice in a variety of empirical or consulting
situations. In the next sections, we use a case study of a major construction project to
illustrate and evaluate these structural properties of communities of practice.
METHODS
This study uses data from Sundlink Contractors, an international contractor consortium
that designed and constructed the Öresund Bridge, a five-mile multi-level bridge
connecting Denmark and Sweden, during 1996-2000. It is important to remark that
the focus of our study represents a highly complex infrastructure project of immense
size, stringent quality requirements, well-defined completion time, and harsh
environmental conditions. In addition, a continuous stream of emergent problems
situations, which are the nexus of CPs, characterized the environment of this research
site. Therefore, this was a very interesting site in which to explore the structural
characteristics of communities of practice.
Sundlink’s organizational structure was functional in nature with personnel
divided into four categories: Operations, Support, Project Management, and Other.
Operations and Support were chosen for this study while Project Management and
Other were excluded due to their small size. Operations included five departments that
were responsible for building the immense concrete and steel structures while the
Support personnel were from two departments and worked with quality and technical
issues. A description of these seven departments is included in Table 1.
The work sites of this project were physically dispersed with only the Quality
and Technical Departments located in the same building. Distances ranging from one
to more than 1,000 kilometers separated the different sites with four of five of the
operational departments located within a ten-kilometer radius of each other. The
various sites corresponded with the different operational entities.
Sundlink Contractors utilized a formal quality system based upon ISO 9000,
which articulated the work processes and procedures. It is within this quality system
that we have defined the joint enterprise for communities of practice in this study
(Schenkel, 2002). In particular, we look at the management of ”deviations” or
situations in which articulated procedures or processes are not followed or articulated
objectives are not achieved. The management of deviations requires 1) the use of
already existing work methods, 2) a change in existing work methods, or 3) the
development of new technical solutions. Thus, examining deviations provides the
context for exploring joint enterprise (management of the deviation within the project),
a shared repertoire (the common means/behavior in which the incidents are

approached and managed), and mutual engagement (the collaboration of multiple
individuals/groups of different and/or the same competences).
Table 1 Description of Departments
Department Activity Size of Structure Location

Offshore To construct prefabricated 800 to 4700 tons Malmö,
Onshore concrete structures called 10 to 51 meters tall Sweden
(Operations) caissons and piershafts
High Bridge To build concrete bridge Over 200 meters tall Malmö,
(Operations) pylons 4355 m3 of concrete Sweden
800 tons of
reinforcement concrete
Bridge Line To construct viaduct using 560 meters long Malmö,
(Operations) steel and concrete Sweden
Prefab Approach To construct steel and concrete 2000 to 6300 tons Cadiz, Spain
Bridge girders 120 meters long
(Operations)
Technical To work with design, survey, Malmö,
(Support) and other technical issues Sweden
QAD To draft and implement a Malmö

(Support) quality system similar to ISO Sweden
9000
A questionnaire was administered during a nine-month period ending in May

1999. Two types of data were collected: 1) communication patterns in managing
deviations, e.g., whom the respondent contacted both within and outside the
organization for advice in situations which deviated from prescribed ISO 9000
standards, and 2) socio-demographic information such as age, education, and
experience data - all potential factors that can influence the formation and maintenance
of communities of practice. The population was delimited by choosing those who
were not construction workers, i.e., those who had a managerial or support function.
In total, 137 people of both an operational as well as support character were included
in the population and 120 people responded (87.6%). Of interest is that only two
respondents provided names that were not within the list of 137 participants. The
mean age of the respondents was 41.29 (s.d. = 11.45), and the mean education was
equivalent to that of a high school diploma. The respondents had worked in the
construction industry a mean of 17.00 years (s.d. = 12.51) and on this particular
project for a mean of 2.15 years (s.d. = 1.11). For a more detailed examination of the
observed communication patterns, see Schenkel & Rognes (1999) as well as Schenkel
(2000).
290 ARTICLE ONE
To illustrate the structural properties of CPs, we performed three separate

analyses to determine to what degree the unit of analysis fulfilled the structural
properties of communities of practice. The three units of analysis were 1) the overall
project based on relationships between individual project members regardless of
department membership, 2) each department based on relationships between the
department individual members, and 3) the overall project based on the aggregated
individual relationships between departments. The data were analyzed using the
UCINET network analysis software package (Borgatti, Everett, & Freeman, 1999) and
imported into Krackplot (Krackhardt, Blythe, & McGrath, 1994), a program used for
the graphical analysis of networks, as well as SPSS.
ANALYSIS AND DISCUSSION
The Overall Project

The first step of our analysis was to look at the overall project structure to determine to
what degree the project as a whole exhibits the structural properties of a community of
practice. We looked at the communication relationships between the individual
project members who were spread across the seven different departments. Figure 1
diagrams the overall project network.
Figure 1 Network of Help-Seeking among All Project Members
Note: Node Shape Indicates Section Membership

Connectedness
Of the 120 project members there were only five isolates or individuals who possessed
zero ties to other members. All other members were connected by at least one tie,
forming a single connected component. This extent of connectivity or connectedness
is consistent with the community of practice structural property 1.
Density
The density of the network was calculated to be 3.9%. Unfortunately, no standard
database of published organizational networks exists that we can use to compare to our
result. However, based on the non-representative sample of the authors’ experience,
we would judge the observed density as quite low for an organizational network of this
size and scope. We have seen that even organizations containing unrelated subunits
(which therefore would not be expected to communicate to a high degree) tend to
achieve a density higher than 3.9%. Hence, if communities of practice were expected
to have an even higher density than typical organizational networks, this would
suggest that this project does not fulfill the second structural property to a high degree.
Graph-theoretic Distance
The average graph-theoretic distance among all pairs of persons in the network
(excluding the five isolates) was 3.551 (s.d. = 1.471). Once again, no standard
database exists for comparison, but experience suggests that this value is certainly no
lower than that obtained in a variety of organizational networks, indicating that the
structure of this network does not fulfill this CP structural dimension to a high degree.
Core/periphery Structure
The crucial structural indicator of a CP is the presence of a core/periphery structure,
i.e., the absence of factions. For this network, we obtained a fit to the core/periphery
model of 0.327, which is significantly greater than zero, but a far cry from what we
conceive as a well-functioning community of practice structure. Thus, this project
does not fulfill the fourth structural property to a high degree either.
Coreness
We then examined the characteristics of those individuals in the core vs. those in the
periphery by correlating coreness with several demographic variables. As shown in
table 2, coreness was not related to age, years of experience in the construction
industry, or years of experience in similar duties elsewhere, but it was significantly
related to the number of years in the current position. In addition, coreness was
negatively related to time spent at the construction site versus at the office where most
people were. Finally, we found that coreness was related to the level of education. A
discussion of how these findings relate to the CP literature is found below.
292 ARTICLE ONE
Table 2 Correlations between Demographics and Corenessª

Variable Coreness
Age -.14
Education .41***
Years in Construction -.07
Years of Similar Duties -.07
Years in Current Position at Sundlink .37***
Time Spent on Construction Site -.44***
ªN ranges from 112-113
* p < .05
** p < .01
*** p < .001
Summary of the Overall Project

Overall, we find very weak support for considering the project as a whole to display
the structural properties of a community of practice. While the project fulfills the first
structural property, it fulfills the next three structural properties to a very low degree.
In particular, there is little support in terms of density that the project network is a
strong community of practice. This low density could be explained by the disciplining
nature of ISO 9000 and its key role in the shaping of practice to the extent that the
formal and informal practices overlapped one another to a high degree in this project,
thus reducing the number of potential communications among individuals (Schenkel,
2002). Furthermore, the geographical separation of sites and uneven resource
allocation meant that not all the people in the respective departments could interact on
a face-to-face basis. This finding supports our argument that face-to-face interactions
are important for the development of communities of practice and that communities of
practice primarily develop within co-located groups of individuals.
The results most consistent with a strong community of practice structure are
those associated with the individual coreness property. Individuals with high coreness
had been in their current position the longest, had higher levels of education, and were
less often at the construction site, all aspects that make sense within the CP literature.
We would expect the relationship with time in current position because the longer a
person participates in a community of practice, the more opportunity there is for
building relationships and moving from the periphery (where all newcomers begin)
into the core. In terms of education, this may be partly due to management’s
preference for hiring more educated people for central positions, requiring a high
degree of theoretical technical knowledge, but it also may be that more educated
people were considered to be more knowledgeable than others and, therefore, more
often approached for help. Alternatively, the coreness of more educated people may
be again that ISO 9000 acts as a disciplining system, i.e., ISO 9000 dictates that
certain individuals should be contacted and in this case these individuals were those
who happened to have a high degree of education (Schenkel, 2002). Finally, the
negative relationship with time at the construction site also makes sense in light of the
CP literature since physical proximity and thus face-to-face communication are

thought to be important facilitators of mutual engagement.
The Departments
Although we find that the project as a whole displays only weak structural properties
of a community of practice, this does not preclude that individual departments might
display strong CP characteristics. Thus, here we look at to what degree an individual
department fulfills the structural properties based on the communication relationships
between the individual members in the department. We look only at the first four
properties, leaving aside the fifth structural property temporarily since it does not
speak directly to the question of to what degree a network shows characteristics of a
community of practice. Due to space constraints, we only look at results from three
departments, chosen because they clearly depict a variety of network attributes.
Technical Department
Visual inspection of the Technical Department in figure 2 shows that its members
were very well connected, with just one isolate. Thus, if we exclude this individual
(35), this department fulfills the first structural CP property. The second property was
fulfilled to a satisfactory degree since this department (among all but the excluded
individual) had an average graph-theoretic distance of 1.91 -- less than two links. The
third property, or department density, is a healthy 27.5%, which is comfortably high.
For the fourth property, the fit of the core/periphery model was 0.569, which is very
high, indicating that this department does have a core/periphery structure, thus
satisfying the fourth property to a high degree. In sum, we find that the Technical
Department as a whole fulfills the first four structural conditions of a community of
practice to a very high degree, thus displaying characteristics of a strong community of
practice.
294 ARTICLE ONE
Figure 2 Technical Department
High Bridge Department

In contrast to the Technical Department, the High Bridge Department is split into two
mutually exclusive subgroups that are largely based on operative sections: circles and
squares, as shown in figure 3. Furthermore, there are four isolated individuals. Due to
the split and presence of isolates, it is not necessary to run the density and graph-
theoretic property analyses since it is clear from the initial analysis that this
department will score extremely low on these. Thus, this department fails to fulfill the
first four structural properties to any kind of satisfactory degree. However, we do see
Figure 3 High Bridge Department

indications of two smaller CPs within the department. These subgroups consist almost
entirely of members from individual sub-department sections (circle and square
nodes). This suggests that interactions within this department are rather strongly
patterned by the formal organizational structure and that the communities of practice
follow the formal organization in this department. An analysis of these two sub-
groups is beyond the scope of this study; however, of additional interest, is that one of
the most central players (31) is in a position of authority (a manager) while the other
central player (34) is a supervisor. Thus, the question of what role formal authority
plays in the formation of CPs comes into question.
Bridge Line Department

Like the previous department, the Bridge Line Department also fails to satisfy the
structural properties for a strong community of practice as a whole. As shown in
figure 4, it is extremely disconnected, divided into many smaller components. Even
the one larger component is nearly disconnected, held together by nodes 89, 124, and
27. Once again, it is interesting to note that some of the key structural nodes that hold
the network together, 124 and 88, are the Quality Controllers.
Figure 4 Bridge Line Department
Summary of the Departments

In summary, only one department fulfills the first four structural characteristics of
communities of practice to a satisfactory degree. However, we did also see indications
of two smaller CPS within the High Bridge Department. Thus, we do see some
296 ARTICLE ONE
support for the relationship between the development of CPs and the formal
organization.
The Overall Inter-department Project

An alternative way of analyzing the project is to determine to what degree the project
at the inter-department level, and not the inter-individual level, satisfies the CP
structural properties. The most common reason why organizations have long graph-
theoretic distances and imperfect fit to the core/periphery model is that they exhibit
departmental homophily – the tendency for individuals within a department to interact
more with members of their own department than with members of other departments.
Normally, strong departmental homophily is inconsistent with the core/periphery
pattern since in effect the departments form a series of cliques. However, if the
departments themselves interact with other departments in a pattern resembling a
core/periphery structure, this will mitigate the effect of cliquing and yield a middle
core/periphery score at the individual level. Hence, we now ask whether the
organization can be seen as a community of practice in which the members are
departments rather than individuals.
To examine this, we constructed an aggregate network at the department level.
We counted the number of ties between each member of each pair of departments and
divided by the maximum number of ties possible. When this ratio exceeded 10% of
possible ties, it was considered as a tie between departments. We refer to this ratio as
the density of ties between departments. Figure 5 shows the resulting network in
which an arrow drawn from one department to another indicates that members of the
first department seek advice from the second department, but not the other way
around. Double-headed arrows indicate reciprocal advice-seeking, with members of
each department seeking advice from members of the other department. The "halo"
atop most departments is a reflexive tie; it indicates that members of that department
go to other members in the same department for assistance.
Figure 5 Network of Inter-department Relations
The results show that the departments are well connected and that there are no isolates.
Thus, the structural property 1 of connectedness is satisfied. The average graph-
theoretic distance in the inter-department network is 1.47 -- less than two links, which
is lower than the expected value for random networks of this size and density.
Therefore, our second property is satisfied to a relatively high degree. The third
property, or density, was found to be 53.06%, which seems to be more than adequate.
The fourth property, the presence of a core/periphery structure, shows that the
departmental network fits extremely well (fit coefficient = 0.69). Thus, the QAD and
Technical Departments form the core, and the Prefab, High Bridge, Onshore, Offshore,
and Bridge Line Departments form the periphery.
Summary of Overall Inter-department

In summary, we can argue that the project at the macro-level fulfills the first four
structural criteria of a community of practice to a considerably satisfactory degree.
The QAD and Technical Departments form the core and are essentially responsible for
the connectedness of the project network as a whole. Without these departments, the
organization would be largely disconnected. This can be seen as underscoring the
importance of such support functions as well as the key role of ISO 9000 in shaping
practice (Schenkel, 2002). However, it should be questioned as to whether the
operational departments should have direct contact with each other instead of through
the QAD and Technical Departments. Thus, this level of analysis provides an
alternative method of understanding how well an organization fulfills the structural
properties of a CP.
298 ARTICLE ONE
Summary of Results
Concluding our analysis and discussion, we find that the inter-department network
displays structural CP characteristics to a higher degree than the project as a whole.
Looking at individual departments, we find that only the Technical Department
exhibits strong structural characteristics. Thus, the suggestion that the shared
repertoire was stronger at the department level than at the project level did not find any
support. In addition, these findings suggest that the formal organization on the
departmental level does not necessarily coincide with emergent communities of
practice, thus confirming previous research on CPs. Finally, the presence of micro-
communities at the sub-department level does suggest that communities do exist below
the entire project and inter-department level.
PROPOSITION DEVELOPMENT
In this section, we now turn to the relationship between CPs and organizational
performance and develop a series of propositions that relate CP structural properties to
performance.
Density and Community Performance

According to structural properties 1 and 3, a certain level of density is required to hold
the CP together. If the CP dips below this, then it is likely that it would disintegrate
and no longer fulfill the structural properties of a CP. But what is to be made of a
community with a density just greater than the minimum that still satisfies the
cognitive aspects of CPs (common enterprise and mutual engagement) and other
structural properties? On the one hand, it seems likely that if knowledge transfer is an
emergent outcome of social interactions -- a fundamental axiom in the CP literature --
then as the density of the network increases so will the transfer of knowledge among
its members. On the other hand, if the density is very high, and particularly so in large
networks, then individuals are spending a considerable amount of time maintaining a
large number of relationships. It has been suggested that community members can
spend too much time managing their relationships to the detriment of their work
(Hansen, 1996). Thus, for smaller communities, we would expect an increase in
density to be associated with an increase in knowledge transfer. In contrast, for larger
communities, we would expect the relationship between density and knowledge
transfer to be curvilinear. Research in non-organizational settings suggests that North
Americans maintain an average of about 20 significant relationships at any given time
(Walker, Wasserman & Wellman, 1994). Thus, in the absence of any other
information, let us assume that this number applies to work relationships as well, and
that more than, say, 40 relationships is clearly difficult to manage. Obviously, the size
of the community is an empirical question that has to be investigated. However, we
suggest 40 individuals as the boundary between smaller and larger communities. We
thus formulate the following propositions:
PROPOSITION 1A: For smaller communities of practice (<40 members),

knowledge transfer increases linearly with density.
PROPOSITION 1B: For larger communities of practice (>40 members, knowledge

transfer increases curvilinearly with density.
Network Centralization, Task Complexity, and Community Performance

Centralization, as we use the term here, refers to the extent to which a network
revolves around a clear core. Operationally, we can measure centralization as a
function of the variance of coreness. If each member has the same coreness value, the
variance is zero, indicating that the network is not centralized at all. If two or three
members have very high coreness values while all the others have very low coreness
values, the variance will be high, indicating strong centralization.
There is a long research tradition, dating back to the 1940s and 1950s (e.g.,
Bavelas, 1948; Leavitt, 1951, etc.), which consistently shows that more centralized
networks do a better job of solving simple problems than less centralized networks do.
Decentralized networks, however, do a better job of solving complex problems.
Similarly, Tushman (1977) reported that high-performing teams working on complex
research projects had more decentralized communication structures than teams
working on more routine projects. Applying these findings to communities of
practice, we propose the following:
PROPOSITION 2A: For communities of practice solving more complex problems,

performance will increase as the variance among members’ coreness values
decreases.
PROPOSITION 2B: For communities of practice solving more routine problems,

performance will increase as the variance among members’ coreness values
increases.
Coreness and Individual Performance

Essential to the notion of CPs is that knowledge and practice are constructed through
interactions among community participants. Members help each other by taking the
time to work through each other’s problems, developing insights into new methods
and new applications for existing knowledge for the community (Wenger, 1998).
Exercising intellect by helping others is also likely to help people maintain and even
improve their own technical skills. Research has shown that individuals with a higher
degree of interaction with other community participants have a higher level of
individual performance (Teigland, 2000). However, peripheral individuals have the
least access to others, both in terms of the number of connections and path lengths.
Therefore, they have less opportunity to gain knowledge from others in the
community, resulting in less community-specific knowledge and a more idiosyncratic
practice. At the same time, their lack of connection with others makes them less
300 ARTICLE ONE
influential and less able to shape the community's practice. This leads us to the
following proposition:
PROPOSITION 3: Community participants with higher coreness scores will have

more community-specific knowledge and thus a higher level of individual
performance.
IMPLICATIONS, LIMITATIONS, AND CONCLUSIONS
Research Implications
Several implications for research were developed during the study. First, it is
important to note that inherent in the concept of communities of practice is the concept
that organizations are homogeneous. This means that a firm’s environment and
evolutionary stages (on a firm level) are not considered. Thus, the introduction of
organizations as heterogeneous entities that face different environments and
evolutionary stages calls into question the concept of a general theory of communities
of practice. We would expect then that the structure of communities of practice should
be affected by the organization’s environment and the nature of the task (Lawrence &
Lorsch, 1967). In adapting this view, we suggest a contingency approach (Galbraith,
1973) toward the structural aspects of communities of practice and that there may be
no one best way of how a community of practice should be structured. Second, the CP
concept does not take fully into account that individual communities evolve and
develop over time. We propose that communities have lifecycles and that the
community may have different characteristics depending upon what point it has
reached in its development. Thirdly, what became evident in the analysis is that the
makeup of the individuals in a community of practice is critical to the structural
aspects of the community and that social network analysis and the concepts we have
developed allow us to analyze CPs. Thus, it would be of interest to further investigate
what constitutes “the right mix” of members of a community in terms of
demographics. Finally, this research has suggested that positions that are not of
importance in the formal organization are actually of high importance in the informal
organization since novices (QCs) were found to be critical in establishing links between
various parts of the organization. This may then suggest that perhaps the best way to
organize is around the emergent informal network. All of these implications deserve
further research, but what is of primary interest is the question of whether or not
management can in fact influence the dynamics of a community of practice since they
are emergent and of an informal nature.
Limitations
We should note the limitations of the study and caution that this study was of an
exploratory nature, with the findings merely acting as guidelines for further research.
We have focused on only one organization, thus limiting the generalizability of our
findings. In addition, for the purpose of this study, we have merely analyzed existing
organizational units to see to what degree they fulfill the structural properties of a
community of practice. However, as research has shown, the boundaries of
communities of practice are ethereal and may not necessarily conform to the formal
boundaries of an organization (Wenger, 1998). Thus, further research needs to look at
how these structural properties may be applied to organizations without being confined
by the formal organizational boundaries.
In addition, since the focus of this study was on the structural dimensions, we
have not taken into consideration any of the cognitive dimensions of CPs when
conducting our analyses. However, further research should look at the interaction
between the structural and cognitive dimensions of communities of practice and their
relationship to organizational performance.
Conclusion
In conclusion, we feel that we have made considerable progress towards our two
objectives, and in the process we have opened the door for further theorizing and
empirical studies. First, our results show that network analysis does provide an
illuminating way to better understand the structural properties of communities of
practice. We have identified specific structural criteria implicit in the logic of
communities of practice, reframed these criteria in terms of social network theory, and
then developed five measurable structural properties (connectedness, graph-theoretic
distance, density, core/periphery structure, and coreness). Using data from a major
complex infrastructure project, we illustrate the usefulness of the structural properties
and find that the project as a whole on an aggregate department level fulfills the
structural CP dimensions to a higher degree than the whole project at an individual
level. In addition, we find that the Technical Department forms the mainstay of the
entire project community's core. Investigating the organization more deeply by
analyzing each individual department, we find that only the Technical Department
satisfies the structural properties for a community of practice to a high degree.
Second, we developed a set of three propositions linking the structural aspects
of a community of practice to performance. Scholars often assume that the more a
community resembles the ideal CP, the better its performance. However, as discussed
above, the connection between performance and communities of practice is a largely
neglected area. Thus, if the CP concept is to be of value, then the connection between
the community and organizational performance must be further investigated. This
research then provides some antecedents with which researchers may examine the
relationship between communities of practice and organizational performance.
REFERENCES
Allen, T.J. 1977. Managing the Flow of Technology: Technology Transfer and the
Dissemination of Technological Information within the R&D Organization.
Cambridge, MA: MIT Press.
Bavelas, A. 1948. A mathematical model for group structure. Human Organizations,
7: 17-30.
302 ARTICLE ONE
Blau, P.M. 1977. Inequality and Heterogeneity: A Primitive Theory of Social

Structure. New York: Free Press.
Bonacich, P. 1972. Factoring and weighting approaches to status scores and clique
identification. Journal of Mathematical Sociology, 2: 113-120.
Boland, R.J. Jr. & Tenkasi, R.V. 1995. Perspective making and perspective taking in
communities of knowing. Organization Science, 6,4: 350-372.
Borgatti, S.P. & Everett, M.G. 1999. Models of core/periphery structures. Social
Networks, 21: 375-395.
Borgatti, S., Everett, M.G., & Freeman, L.C. 1999. UCINET 5 for windows. Natick,
MA: Analytic Technologies.
Brown, J.S. & Duguid, P. 1991. Organizational learning and communities of practice.
Conner, K.R. & Prahalad, C.K. 1996. A resource-based theory of the firm: Knowledge
versus opportunism. Organization Science, 7: 477-501.
Degenne, A. & Forsé, M. 1999. Introducing Social Networks. London: Sage.
Drucker, P. 1991. Post-capitalist Society. Oxford: Butterworth-Heinemann.
Everett, M.G. & Borgatti, S.P. 1999. Peripheries of cohesive subsets. Social Networks,
21: 397-407.
Friedkin, N. 1982. Information flow through strong and weak ties in
interorganizational social networks. Social Networks, 3: 273-285.
Harary, F. 1969. Graph Theory. Reading, MA: Addison-Wesley.
Kanter, R.M. 1983. The Change Masters. New York: Simon & Schuster.
Kanter, R.M. 1989. The new managerial work. Harvard Business Review, 6: 85-92.
Kogut, B. & Zander, U. 1992. Knowledge of the firm, combinative capabilities and the
replication of technology. Organization Science, 3,3: 383-397.
Kogut, B. & Zander, U. 1995. Knowledge of the firm and evolutionary theory of the
multinational corporation. Journal of International Business Studies, 26: 625-644.
Kotter, J.P. 1982. The General Managers. New York: Free Press.
Kotter, J.P. 1985. Power and Influence: Beyond Formal Authority. New York: Free
Press.
Krackhardt, D., Blythe, J., & McGrath, C. 1994. Krackplot 3.0: An improved network
drawing program. Connections, 17, 2: 53-55.
Lave, J. & Wenger, E. 1991. Situated Learning: Legitimate Peripheral Participation.
Lawrence, P.R., & Lorsch, J.W. 1967. Organization and Environment: Managing
Differentiation and Integration. Boston: HBS Press.
Leavitt, H.J. 1951. Some effects of communication patterns on group performance.
Journal of Abnormal and Social Psychology, 46: 38-50.
Liebeskind, J.P. 1996. Knowledge, strategy and theories of the firm. Strategic
Management Journal, 17: 45-62.
Liedtka, J. 1999. Linking competitive advantage with communities of practice.
Journal of Management Inquiry, 8: 5-16.
Miles, R. & Snow, C. 1994. Fit Failure and the Hall of Fame. NY: Free Press.
Mitchell, J. C. 1969. Social Networks in Urban Situations. Manchester, England:
Manchester University Press.
Orr, J. 1990. Talking About Machines: An Ethnography of a Modern Job. Ithaca, NY:
Cornell University.
Prahalad, C.K. & Hamel, G. 1990. The core competence of the corporation. Harvard
Business Review, 68: 79-91.
Ranson, S., Hinings B., & Greenwood, R. 1980. The structuring of organizational
structures. Administrative Science Quarterly, 25: 1-17.
Rogers, E.M. 1995. Diffusion of Innovations. New York, NY: The Free Press.
Schenkel, A. 2002a. Communication in incidents at the Öresund bridge. Paper
presented at the Construction Economics and Organization: Nordic Seminar,
Gothenburg, Sweden.
Schenkel, A. 2002b. Conceptualizing and exploring the second level effects of ISO
9000. Presented at Third European Conference on OKLC 2002, Athens, Greece.
Simon, H.A. 1945. Administrative Behavior. New York: The Free Press.
Spender, J.-C. 1989. Industry Recipes: The Nature and Sources of Managerial
Judgement. Oxford: Basil Blackwell.
Spender, J.-C. 1996. Making knowledge the basis of a dynamic theory of the firm.
Strategic Management Journal, 17: 45-62.
Spender, J.-C., & Grant, R.M. 1996. Knowledge and the firm: Overview. Strategic
Storck, J. & Hill, P.A. 2000. Knowledge and diffusion through “strategic
communities”. Sloan Management Review, Winter: 63-74.
Teigland, R. 2000. Communities of practice in an internet firm: Netovation vs. on-time
performance. In E. Lesser (ed.), Knowledge and Communities. Newton, MA:
Butterworth-Heinemann.
Tushman, M. 1977. Special boundary roles in the innovation process. Administrative
Walker, M., Wasserman, S., & Wellman, B. 1994. Statistical models for social support
networks. In S. Wasserman & J. Galaskiewicz (eds.), Advances in Social Network
Analysis. London: Sage.
Wasserman, S. & Faust, K. 1994. Social Network Analysis: Methods and Applications.
Wenger, E. 1998. Communities of Practice: Learning, Meaning, and Identity.
Article Two
The Provision of Online Public Goods: Examining Social

Structure in an Electronic Network of Practice
In general, the terminology in this article corresponds to the terminology in this

thesis. We use one abbreviation:
1) ENOP for Electronic Network of Practice

306 ARTICLE TWO
Previous version published in the Proceedings of the 23rd International Conference on

Information Systems (ICIS), Barcelona, Spain, 2002
The Provision of Online Public Goods:

Examining Social Structure in an Electronic Network of
Practice
Molly McLure Wasko

MIS, College of Business, Florida State University, Tallahassee, FL, USA
email: mwasko@cob.fsu.edu
Robin Teigland
Institute of International Business, Stockholm School of Economics, Stockholm,
Sweden
ABSTRACT
Electronic networks of practice (ENOP) exist primarily through computer-mediated
exchange and are social spaces where individuals working on similar problems self-
organize to help each other and share perspectives about their occupational practice or
common interests. This exchange of knowledge through message postings produces
an online public good, where all participants in the network can access the knowledge,
regardless of their contribution. Thus, this research builds upon theories of collective
action and public goods to better understand the provision and maintenance of
knowledge in an electronic network of practice. We use social network analysis to
examine the following research questions: 1) what is the pattern of contribution that
produces and sustains the ENOP public good, 2) are ENOPs characterized by a critical
mass constituting a core? and 3) how does the heterogeneity of resources and interests
of participants impact ENOP collective action? We find that the network of practice is
sustained through generalized exchange, is supported by a critical mass of active
members, and is shaped as a star. The critical mass is significantly related to
occupational tenure, expertise, availability of local resources and a desire to enhance
one’s reputation.
Keywords: collective action, public goods, knowledge, electronic community,

community of practice, social network analysis, law
PROVISION OF ONLINE PUBLIC GOODS 307
INTRODUCTION
Recent advances in information and communication technologies (ICT) have led to the
development of extra-organizational electronic networks. These electronic networks
enable the creation of weak structural links between thousands of geographically
dispersed individuals, who are typically strangers and come from diverse
organizational, national, and demographic backgrounds (Sproull & Faraj, 1995). In
these networks, individuals are able to engage in knowledge sharing, problem-solving,
and learning through posting and responding to questions on professional advice,
storytelling of personal experiences, and debate on issues relevant to the community
(Wasko & Faraj, 2000). Individuals benefit from these networks since they gain
access to new information, expertise, and ideas that are often not available locally.
Termed electronic networks of practice (ENOP), these electronic networks exist
primarily online and are similar to communities of practice in that they are a social
space where individuals working on similar problems self-organize to help each other
and share perspectives about their occupational practice or common interests (Brown
& Duguid, 2000).
Despite the growing interest in online cooperation and virtual organizing,
surprisingly little empirical research has investigated the communication and
organizing processes in online networks (Monge et al., 1998; Lin, 2001). Thus, the
goal of this research is to better understand these emerging organizational forms by
drawing upon the well-established theories of public goods and collective action.
Building upon work by Fulk and colleagues (Fulk, Flanagin, Kalman, Monge, & Ryan,
1996), we extend collective action theories to interactive communication systems,
examining participation in ENOPs as a form of collective action. The collective action
is exhibited through the interactive posting and responding of messages to the
network. This interaction produces and maintains the public good of a continuous
stream of relevant practice knowledge that all participants may access.
This extension of collective action and public goods theories to ENOPs is
highly exploratory. Thus, rather than test hypotheses, we develop and examine three
research questions that we see as fundamental to understanding ENOPs: 1) what is the
pattern of contribution that produces and sustains the ENOP public good, 2) are
ENOPs characterized by a critical mass constituting a core? and 3) how does the
heterogeneity of resources and interests of participants impact ENOP collective
action? To address these questions, we collected postings from a successful ENOP
during two months and then administered a survey to all active participants. The
shared practice of this network was US federal law, where participants (lawyers)
actively engaged in exchanging legal advice. The paper concludes with a discussion
of findings and areas for future research.
308 ARTICLE TWO
THEORY DEVELOPMENT
Members of a collective must often make decisions that balance the benefits of
maximizing self-interest with the collective’s interests. This phenomenon is best
explained through an example, such as wheat farming. Wheat prices are primarily set
by market supply and demand. Each farmer attempts to grow and sell as much wheat
as possible to obtain the greatest profit. However, in surplus years there is the risk that
prices fall dramatically if farmers flood the market with wheat. Thus, for the
collective, the rational action is to restrict the total market supply of wheat to maintain
high prices. However, each farmer has the individual incentive to try to sell as much
wheat as possible. Yet when each individual acts rationally, the market is flooded and
wheat prices fall, leaving everyone with less profit. Thus, the sum of individually
rational actions leads to collective irrationality, leaving everyone worse off.
This example is a special problem referred to as a social dilemma. Social
dilemmas arise when a set of individuals act rationally in their own self-interest, yet
the sum of their actions leads to collective irrationality (Kollock, 1998). Social
dilemmas involving more than two individuals are N-person dilemmas and fall into
two categories, the provision of public goods and the tragedy of the commons. First,
the provision of public goods dilemma, or the social fence, involves the production of
a public good. Public goods are resources from which all individuals in a collective
may benefit regardless of whether they have contributed to providing the good, such as
a public park or public television (Kollock, 1998). In the provision of public goods
dilemma, the optimal individual decision is to enjoy the public good without
contributing anything to its creation or maintenance.
The specific characteristics of a public good have implications for its provision
and use. First, a public good is a resource that can be provided only if members of a
collective contribute towards its provision. It is non-excludable, i.e., the good cannot
be withheld from any member of the collective, even if he or she does not participate
in the production or maintenance of the good (Komorita & Parks, 1995). A second
characteristic is known as non-rival, meaning that the good is not used up or depleted
in its consumption, thus one person’s use of the good does not diminish its availability
to others in the collective (Shmanske, 1991). Public goods are generally considered to
evidence both non-rivalry and non-excludability. Since public goods are not used up
in their consumption due to non-rivalry, there is no incentive to add costs by
controlling access to the good through exclusion (Musgrave, 1959). However, a
connection between the two characteristics of non-rivalry and non-excludability does
not necessarily exist: a non-rival good can be excludable while a non-excludable good
can be either rival or non-rival (Shmanske, 1991). Thus, true public goods are
completely non-excludable and non-rival; however, it is argued that many public
goods exhibit these characteristics to varying degrees (Kollock, 1998).
The second type of social dilemmas is the social trap or the tragedy of the
commons and involves the consumption or replenishment of a joint good. The
commons dilemma differs from the provision of public goods dilemma because the
joint good is not a public good. Rather the joint good is subtractable, the opposite of
non-rival. In other words, the use of the joint good by one individual diminishes the
availability of the good to another individual, resulting in the “tragedy of the
commons” (Kollock & Smith 1996).
This research focuses on the production of knowledge as a public good in
ENOPs. In the formal language of collective action theory, the network participants
are the interest group and the public good is the continuous stream of knowledge
produced and jointly held by the network’s participants. We argue that the knowledge
produced by the ENOP in this study is a true public good. First, it is non-excludable
due to the network’s open nature. When one participant responds to a posting, then all
members may benefit from this knowledge, even though they did not contribute to its
production through either posting or responding. Second, the knowledge is non-rival
because even if the person receiving the help uses the knowledge, it still remains
available to other members, who may also apply the knowledge in their own
situations. Thus the social dilemma faced in ENOPs is the provision of public goods
dilemma. Specifically, we are interested in understanding how open, voluntary
ENOPs are sustained, given that individuals are better off not contributing and free-
riding on the efforts of others.
Prior research in social dilemmas has identified critical factors underlying the
production and maintenance of public goods (Oliver, Marwell, & Teixeira, 1985;
Oliver & Marwell, 1988; Fulk et al., 1996). Building upon this research, we examine
these factors through three research questions, providing evidence from our study.
Specifically, we examine the patterns of exchange, the existence of critical mass, and
the heterogeneity among actors to better understand the provision of online public
goods in networks of practice.
STUDY DESIGN AND DATA COLLECTION

Conducted in a field setting, this study examines a single extra-organizational ENOP
of a US professional legal association. All association members have ENOP access as
part of their membership benefits, yet participation is voluntary. Individuals were
chosen to take part in this study based on their ENOP participation, which consisted of
posting a message to the network during the two months of April and May 2001. This
ENOP is supported by “bulletin board” technology, similar to that of Usenet
newsgroups where questions and responses are connected in a “thread”, resembling a
conversation.
The name of the person posting was included in each message. During the two
months, there were 2,460 messages posted to the network by 526 unique individuals.
Each participant was sent a survey and we received 152 valid responses for a response
rate of 29%. To assess response bias, we compared the participation rates of survey
responders with those of non-responders and found that the participation rates of the
two groups were not significantly different (F = .823, ns). We use both the objectively
collected message postings as well as survey results to examine our research questions.
The following sections examine the three research questions as well as the specific
data and methods used to explore each question.
310 ARTICLE TWO
RESEARCH QUESTIONS AND RESULTS
RQ1: What is the Pattern of Contribution That Produces and Sustains the
ENOP Public Good?
The first key issue for examination is the pattern of contributions that create the good.
In ENOPs, contribution is reflected in the posting of questions and replies that take the
form of a conversation. This interaction creates social ties between participants. We
define a social tie in ENOPs as the tie created between two individuals when one
person responds to another’s posting. While it has been argued that social ties are
important for collective action, it is less well established as to exactly how and why
social ties are important (Marwell & Oliver, 1988). Initial research proposes that the
overall frequency or density of social ties within a group is related to the achievement
of collective action. When networks are dense, consisting of direct ties between all
members, collective action is relatively easier to achieve. This argument goes back to
Marx, who reasoned that the more individuals are in regular contact with one another,
the more likely they will develop a “habit of cooperation” and thus act collectively
(Marwell et al., 1988). Thus, one view is that ENOPs may be characterized by a dense
network structure, where all members interact with all other members.
An alternative view suggests that the pattern is more like a reciprocal gift
exchange. This view suggests there is a dyadic exchange between a help provider and
a help seeker, with the expectation that the gift of help will be reciprocated some time
in the future (Kollock, 1999). Thus, the nature of exchange in an ENOP may be
structured as reciprocal dyadic exchanges between individuals, where the motivation
to help others stems from the expectation of obligation and reciprocity from the
receiver. A third view stemming from collective action argues that public goods can
be provided through generalized exchange (Fulk et al., 1996). A generalized exchange
occurs when one’s giving is not reciprocated by the recipient, but by a third party
(Ekeh, 1974). In contrast to dyadic exchange characterized by direct reciprocity and
accountability, generalized exchange is based on indirect reciprocation and interest-
based contribution. Therefore, ENOPs may also be sustained through generalized
exchange.
RQ1 Results. All ENOP messages were examined to determine the identity of the
person posting, and were then coded as seeds (the first message in a thread), singletons
(seeds without responses), questions, responses, or other. We built a social network
matrix consisting of all 526 participants to determine who was responding to whom,
creating a directional, social tie. The first question examines whether individuals are
participating equally. If people participate equally, we would expect all participants to
have posted the average of 4.7 messages. However, the median participation is two
message postings, and 64.8% of network members posted less than four messages,
indicating that people are not equally sustaining the public good. The frequency rates
of participation are provided in table 1.
Table 1 Frequency of Participation

Messages Cumulative
Individuals Percent
Posted Percent
1 173 32.89 32.89
2 109 20.72 53.61
3 59 11.22 64.83
4 34 6.46 71.29
5 28 5.32 76.62
6-10 72 13.68 90.30
>10 51 9.69 100.00
We are also interested in examining whether the exchanges are directly

reciprocal or generalized. The data indicate that there were 1306 exchanges between
455 individuals, and 130 were reciprocated by the same individuals. Thus, only 10%
of the contributions sustaining the network are directly reciprocal. General findings
and results are summarized in table 2. These findings indicate that people do not
participate equally in the public good provision, rather the ENOP is sustained through
generalized exchange between members.
Table 2 Summary of Exchanges

Exchange Characteristics Result
Number of Unique Participants 526
Number of Messages Posted 2,460
Average Participation Rate 4.7 messages / person
Number of Seeds 1,121 by 436 individuals
Number of Singletons 104 by 71 individuals
Number of Threads 1,017, average length 2.4 messages
Dyadic Exchanges 1,306 between 455 individuals
Unique Exchanges 1,176
Direct Reciprocal Exchanges 130, 10% of total
RQ2: Are ENOPs Characterized by a Critical Mass Constituting a Core?

Until recently, the majority of formal collective action analyses assumed that
contribution was uniformly distributed across members. However, RQ1 results
indicate that this is not so in ENOPs. A competing argument suggests that a subset of
the group may be responsible for making the majority of the contributions to the
production and maintenance of the public good. Borrowing from nuclear physics, this
subset is labeled critical mass, referring to the idea that a certain threshold of
participation or action has to form before a social movement may come to exist (Oliver
& Marwell, 1988). As in RQ1, this property is examined by looking at the pattern of
social ties. The presence of critical mass is determined by the degree to which ties are
centralized or concentrated to a few individuals rather than spread across the entire
group.
312 ARTICLE TWO
In order to investigate the critical mass principle in our ENOP setting, we build upon
some of the ideas underlying communities of practice (COPs). Wenger (1998) has
suggested that there are different levels of COP participation: 1) full participation
(insider), 2) peripherality (legitimate peripheral participant or LPP), and 3) full non-
participation (outsider). In full participation, the person is an inclusive community
member. He or she has gained legitimacy through engaging with other community
actors in common actions and has acquired the formal and informal ability to behave
as a community member (Lave & Wenger, 1991). Peripherality connotes legitimate
partial participation in the community. Full non-participation is total exclusion from
the community and occurs because the individual either does not desire to participate
or the individual is not allowed to participate by the community.
Further, it is argued that a true COP has a central group of insiders to whom all
other LPPs are connected (Wenger, 1998). If a community of individuals is divided
into cliques or factions with separate central groups of active individuals then this is
effectively a number of separate communities or constellations that are loosely
connected to each other (Wenger 1998). Thus, we are interested in examining whether
the ENOP has a critical mass of participants sustaining the good for all as well as
whether the ENOP consists of multiple constellations of loosely related cliques.
RQ2 Results. We investigate the question of critical mass by examining the

participant matrix. As apparent from table 1, some members are significantly more
active than others, indicating the likelihood of a critical mass. Building upon
Wenger’s categories and based upon the analysis of messages, we created four
categories of participants: outsiders (people who posted seeds, but never received a
response), seekers (people who posted only questions), periphery (people who posted
10 or less responses) and insiders (people who posted more than 10 responses). Using
UCINET software (Borgatti, Everett, & Freeman, 1999), we analyzed the ego network
of each individual to determine centrality in terms of “in degree” (the number of times
other people respond to an individual) and “out degree” (the number of times an
individual responds to others). Table 3 summarizes these categories.
Table 3 Categorization of Participation

Average Range Total Average Range Total
Category # In In In Out Out Out
Degree Degree Degree Degree Degree Degree
Outsiders 71 0 n/a 0 n/a 0
Seekers 166 2.08 1-15 346 0 n/a 0
Periphery 266 2.53 1-15 673 2.48 1-10 660
Insiders 23 12.48 4-33 287 28.09 11-114 646
This analysis indicates that there is a critical mass of 23 insiders who are the
most active in posting responses to other members. The second analysis examines the
extent to which the critical mass is a clique (responding only to each other) and draws
upon recent work extending social network analysis to COPs (Schenkel, Teigland, &
Borgatti 2002). First, we analyzed the data matrix using UCINET to determine
whether the network has a core/periphery structure. A low core/periphery score of
0.27 indicates that there is no central core of individuals closely tied to each other.
Second, we performed a component analysis, which revealed that the ENOP is
characterized by only one component and not a set of subsets, indicating that the
ENOP does not have multiple cliques.
Additional analysis indicates that the 4% of members who comprise the insider
critical mass posted 646 responses (50% of exchanges). Of these responses, 84% were
to unique individuals, also indicating little overlap within the critical mass. In
addition, this analysis indicates that peripheral members are actively engaged in
generalized reciprocity: LPPs are the recipients of 52% of the exchanges and are
responsible for sustaining 50% of the exchanges by replying to others. By analyzing
the unique number of participants in these exchanges, we note that of the 673
messages received, 614 (91%) were from unique individuals. Of the 660 response
messages posted to others, 635 were to unique individuals (96%).
Thus, the network is structured as a star with a critical mass surrounded by
peripheral connections emanating outwards. There are no cliques, rather the critical
mass actively responds to many unique and overlapping individuals, and the periphery
engages in both receiving and providing advice to others. Using Krackplot, Figure 1
shows the network structure of survey respondents (Krackhardt, Blythe, & McGrath,
1994).
Figure 1 Network Structure

314 ARTICLE TWO
RQ3: How Does the Heterogeneity of Resources and Interests of

Participants Impact ENOP Collective Action?
A population’s heterogeneity of resources and interests are argued to affect collective
action (Olson, 1965; Hardin, 1982; Oliver et al., 1985). The more heterogeneous a
group is, the more likely there is a critical mass or subset of members who have a high
enough level of resources and/or interests to produce the public good. However,
heterogeneity can also hinder collective action even when the mean levels of
heterogeneity appear sufficient. As such, the distribution of heterogeneity is important
in terms of collective action, i.e., the more positive skew and deviation from the mean,
the more likely a critical mass may result (Oliver et al., 1985).
Looking at interests, in most collectives, individuals have differing levels.
Individuals with higher interest levels are those who tend to gain more from additional
contributions to the public good. Hardin (1982) argued that individuals with a high
interest level are those who lack private alternatives. Individual interests also vary in
their underlying motivations for seeing the good realized, which also affects the
potential contribution level (Marwell et al., 1988). These motivations may include
social and/or professional motivations (Wasko et al., 2000).
Resources include money, time, expertise, energy, and influence (Oliver et al.,
1985). For a public good to be produced and maintained, it is argued that those
forming the critical mass are more likely to have access to the required resources.
Previous research has found that people who have higher levels of professional
expertise and organizational tenure are more likely to provide useful advice on
computer networks (Constant, Sproull, & Kiesler, 1996). Thus, the final research
question examines the role of individual interests and resources underlying the
provision of online public goods.
RQ3 Results. We examine the importance of resources and interests by analyzing the
correlations between network centrality data and survey measures of resources and
interests. The survey assessed two types of resources: 1) ENOP expertise measured by
the number of months of professional association membership (objective measure
from association member database) and 2) professional expertise measured by self-
rated expertise. We assessed four types of interests: 1-2) professional motivations of
reputation, and a desire to learn and challenge oneself, 3) social motivation of
sustainability of participation, and 4) lack of private alternatives. Alternatives were
assessed by examining the type of law firm (sole practitioner = 1, associate = 2,
partner = 3), indicating that a lawyer in a sole partnership would have fewer private
alternatives for professional discussion than a lawyer in a law firm with more
colleagues. Reliabilities and validity of constructs demonstrated convergent and
discriminant validity. The multi-item constructs were calculated by taking the average
of the items. Actual items, reliabilities and factor analysis are reported in Appendix A.
Table 4 presents the correlations between constructs.
This analysis suggests that the resources and interests examined in this study
had little correlation with people receiving help (in degree). The only significant
relationships with in degree are sustainability and challenge, thus those who receive
help are interested in continuing their ENOP participation and the challenge associated
with doing so. Resources and interests had higher associations with responding to
others (out degree). The results indicate that longer professional association tenure
and higher levels of expertise are associated with responding to others. In addition,
individuals who are sole practitioners are significantly related to responding to others
as are those concerned with enhancing their reputations. Thus, while interests and
resources were not as significant for people who receive help, they are reasonably
good indicators of why people provide knowledge to others.
Table 4 Correlations of Constructs

1 2 3 4 5 6 7
1 Months in Assoc
2 Expertise .44**
3 Type of Firm .16* .01
4 Reputation .04 -.01 .05
5 Sustainability -.40** -.26** -.04 .25**
6 Challenge -.39** -.23** -.10 .16* .68**
7 In Degree -.01 .03 -.09 .12 .23** .15*
8 Out Degree .17** .15* -.15* .18* .12 .02 .73**
* p < .05
** p < .01
DISCUSSION AND AREAS FOR FURTHER RESEARCH

Theories of collective action and public goods contribute significantly to our
understanding of ENOPs. In this particular ENOP, the public good of knowledge was
produced through a generalized exchange between members. However, this exchange
was not evenly conducted by all members, rather it was sustained by a critical mass of
individuals who primarily responded to others and not only to themselves. This
critical mass was then surrounded by a group of peripheral individuals who both asked
and received advice. Thus, the network is structured as a star with a central critical
mass and connections radiating outwards. In addition, the heterogeneity of resources
and interests provided good indications of why people contributed to the public good
provision. Therefore, we have support to proceed further with these theories to help us
understand ENOP dynamics.
However, we examined only one specific type of ENOP, an extra-
organizational network using bulletin board technology. Other types of ENOP
interactive technology exist such as listservs, chatrooms, and voice, and the use of
these different media may affect ENOP dynamics. In addition, this study was
conducted over two-months and relied on cross-sectional survey data. Thus, we were
not able to investigate changes over time or how the public good was achieved in the
316 ARTICLE TWO
first place. Subsequent studies should be longitudinal to understand ENOP lifecycles.

Longitudinal analysis should also look at the nature of interdependence of individuals’
decisions to contribute to the public good. It has been argued that reciprocal
interdependence and not sequential interdependence characterizes interactive
communication systems (Fulk et al., 1996). However, it has yet to be tested
empirically.
A final issue of interest to managers and researchers is the problem of free-
riders and how they affect ENOP dynamics. Free-riders are those “who do not
contribute sufficiently to the jointly held body of information while continuing to
enjoy its benefits” (Fulk et al., 1996: 78). Two explanations have been provided: 1)
greed or the desire to obtain the best possible outcome for oneself and 2) the “fear of
being a sucker” or the fear that no one else will contribute even though one wants to
(Kollock, 1988:189). In ENOPs, individuals may free-ride through lurking, reading all
messages to gain access to the network’s knowledge without ever posting themselves.
There is also the question of whether people who continually ask questions, receive
help from the ENOP, but never bother to help anyone else in the ENOP are free-riders.
It can be argued that these individuals actually do contribute to the public good
because they stimulate a thought process by other participants. However, this
participation only works if there is a critical mass of individuals who continue to
respond to postings. Finally, while a participant may be interested in contributing, if
there is low ENOP activity, then he or she may feel that their actions will not be
reciprocated the next time they need help and thus, their time spent helping is lost.
In conclusion, this study’s goal was to apply the theoretical lens of collective
action and public goods to examine online cooperation through the provision and
maintenance of knowledge in ENOPs. Our findings suggest some practical
implications for the development and maintenance of ENOPs. First, ENOPs do not
need equal member participation, but rather can be sustained through the collective
actions of a small percentage of members who form a critical mass. This critical mass
is able to provide the public good through generalized exchange of advice and
solutions. These individuals are concerned with enhancing their reputations in the
network, thus technology that supports identifiers of individuals will more likely
succeed than systems where participation is anonymous. In addition, those most likely
to develop the critical mass are tenured experts in their area, but do not have easy
access to interested others. Thus, unlike COPs that require face-to face interaction,
ENOPs transcend traditional barriers to knowledge exchange through the creation of
knowledge as a communal public good, available to all members of the collective.
REFERENCES
Borgatti, S., Everett, M. G., & Freeman, L. C. 1999. UCINET 5 for Windows. Natick,
MA: Analytic Technologies.
Brown, J. S. & Duguid, P. 2000. The Social Life of Information. Boston, MA: Harvard
Constant, D., Sproull, L., & Kiesler, S. 1996. The kindness of strangers: The
usefulness of electronic weak ties for technical advice. Organization Science, 7, 2:
119-135.
Ekeh, P.P. 1974. Social Exchange Theory: The Two Traditions. London: Heinemann
Educational Books.
Fulk, J., Flanagin, A. J., Kalman, M. E., Monge, P. R., & Ryan, T. 1996. Connective
and communal public goods in interactive communication systems. Communication
Theory, 6,1: 60-87.
Hardin, R. 1982. Collective Action. Baltimore, MD: Johns Hopkins University Press.
Kollock, P. & Smith, M. A. 1996. Managing the virtual commons: Cooperation and
conflict in computer communities. In S. Herring (ed.), Computer-Mediated
Communication: Linguistic, Social and Cross-Cultural Perspectives. Amsterdam:
John Benjamins: 109-128.
Kollock, P. 1998. Social dilemmas: The anatomy of cooperation. Annual Review of
Sociology, 24: 183-214.
Kollock, P. 1999. The economies of online cooperation: Gifts, and public goods in
cyberspace. In M. A. Smith & P. Kollock (eds.), Communities in Cyberspace.
London: Routledge: 220-239.
Komorita, S. S. & Parks, C. D. 1995. Interpersonal relations: Mixed motive
interaction. Annual Review of Psychology, 46: 183-207.
Krackhardt, D., Blythe, J., & McGrath, C. 1994. 3.0: An improved network drawing
program. Connections, 17, 2: 53-55.
Lave, J. & Wenger, E. 1991. Situated Learning. Legitimate Peripheral Participation.
Cambridge, UK: Cambridge University Press.
Lin, N. 2001. Social Capital. Cambridge, UK: Cambridge University Press.
Marwell, G. & Oliver, P. 1988. Social networks and collective action: A theory of the
critical mass III. American Journal of Sociology, 94,3: 502-534.
Monge, P. R., Fulk, J., Kalman, M. E., Flanigan, A. J., Parnassa, C., & Rumsey, S.
1998. Production of collective action in alliance-based interorganizational
communication and information systems. Organization Science, 9,3: 411-433.
Musgrave, R.A. 1959. The Theory of Public Finance. New York: McGraw-Hill.
Oliver, P. & Marwell, G. 1988. The paradox of group size in collective action: A
theory of the critical mass II. American Sociological Review, 53: 1-8.
Oliver, P. E., Marwell, G., & Teixeira, R. 1985. A theory of critical mass I: Group
heterogeneity, interdependence and the production of collective goods. American
Journal of Sociology, 91: 522-556.
Olson, M. 1965. The Logic of Collective Action. Cambridge, MA: Harvard University
Press.
Samuelson, P. A. 1954. The pure theory of public expenditure. Review of Economics
and Statistics, 36,4: 387-389.
Schenkel, A., Teigland, R., & Borgatti, S. 2002. Theorizing structural dimensions of
communities of practice: A social network approach. Presented at Academy of
Management.
Shmanske, S. 1991. Public Goods, Mixed Goods, and Monopolistic Competition.
College Station: Texas A&M University Press.
318 ARTICLE TWO
technology. In B. K. J. Keller (ed.), Public Access to the Internet. Cambridge, MA:
MIT Press: 62-81.
Wasko, M. & Faraj, S. 2000. It is what one does: Why people participate and help
others in electronic communities of practice. Journal of Strategic Information
Systems, 9,2-3: 155-173.
Wenger, E. 1998. Communities of Practice. Cambridge, UK: Cambridge University
Press.
APPENDIX A
Į 1 2 3 4 5 6
Number of Months in Association 0.88 0.27 0.09 0.05 -0.25 -0.13
Self-rated Expertise 0.22 0.94 0.00 -0.02 -0.13 -0.10
Type of Practice 0.07 0.00 0.99 0.06 -0.02 -0.03
I earn respect from others by participating on the ENOP .87 0.00 0.18 -0.04 0.87 0.06 0.11
-
I feel that participation improves my status in the profession 0.09
-0.04 0.07 0.91 0.10 0.00
Participating on the ENOP improves my reputation in the
0.13 -0.16 0.05 0.85 0.11 0.05
profession
-
I intend to continue participating on the ENOP .83
0.15
-0.03 0.03 0.15 0.79 0.31
-
I intend to use the ENOP for the foreseeable future 0.11
-0.08 -0.01 0.09 0.82 0.40
-
I intend to use the ENOP at least as regularly as I do now 0.09
-0.12 -0.04 0.10 0.91 0.14
Participating on the ENOP gives me the opportunity to learn new -
.88 -0.12 -0.12 0.01 0.44 0.69
things 0.33
I participate on the ENOP to be exposed to complex problems and
0.12 -0.10 -0.03 0.08 0.21 0.89
issues
-
I find participating on the ENOP interesting 0.31
0.05 0.07 0.08 0.42 0.72
Article Three
Communities of Practice in a High-Growth Internet

Consultancy: Netovation64 vs. On-Time Performance
We use the term ‘community of practice’ to represent networks of practice both

within the firm and outside the firm. Additionally, we use the term on-time
performance to indicate efficient performance.
64 Netovation has been used to describe the use of the internet as a source of creativity or innovation.
320 ARTICLE THREE
In E. L. Lesser, M.A. Fontaine, & J.A. Slusher (eds.), Knowledge and Communities,
Boston: Butterworth-Heinemann, 2000: 151-178.
In J. Birkinshaw & P. Hagström (eds.), The Flexible Firm, London: Oxford University
Press, 2000: 126-146.
In F. Delmar & P. Davidsson (eds.), Tillväxtföretagen (High-Growth Firms),

Stockholm: SNS Förlag, 2001 (in Swedish).
Communities of Practice in a High-Growth Internet

Consultancy: Netovation65 vs. On-Time Performance
Robin Teigland
Sweden
ABSTRACT
This article describes the findings from a study of the patterns of individual-level
knowledge flows at Icon Medialab, a high-growth internet consultancy with
multinational operations, and the impact of those patterns on individual performance.
Building on the knowledge-based view of the firm literature, and specifically the work
concerned with communities of practice, a series of propositions linking various
sources of knowledge (internal vs. external, tacit vs. codified) to individual
performance are developed. Using data collected from 203 employees at Icon
Medialab, it is found that creativity is associated with social contact and internet-based
sources such as electronic communities, while on-time delivery of results is associated
with the use of codified internal sources and negatively related to the use of internet-
based sources. Implications for community of practice theory, and for practice, are
discussed.
Keywords: community of practice, knowledge, internet, performance
65 Netovation has been used to describe the use of the internet as a source of creativity or innovation.
COMMUNITIES OF PRACTICE IN A HIGH GROWTH CONSULTANCY 321
INTRODUCTION
Seldom does a day go by in which one does not read about the knowledge-based
economy. New business models as well as whole new industries are popping up,
resulting in an increasing number of high-growth firms. Yet, despite all the publicity
and information about these firms, we are still a long way from understanding the
workings of these new high-growth firms as well as the knowledge-based economy.
Many of the management theories that we have at our disposal today were developed
in the pre-internet era and may no longer be applicable. For example, we have a very
limited understanding of how individuals in these new age firms exchange knowledge
or how the internet is affecting knowledge flows across firm boundaries. In addition
to the interest in these high–growth, internet-based firms is the rapidly growing
interest in the relatively new field of communities of practice by both academics and
practitioners. In numerous firms, management is attempting to support or formalize
these informal organizational forms in the hope of improving the firm’s competitive
advantage based on knowledge (Boland & Tenkasi 1995, Brown & Duguid 1998,
Davenport & Prusak 1998). Thus, the purpose of this article is to tie the two areas
above together by focusing on communities of practice within the setting of a high-
growth, internet-based firm. The primary intent is to understand how individuals
access knowledge in their everyday work in an internet-intensive environment and
what role the internet and communities play. The second intent is to then take this
research one step further by linking an individual’s knowledge access behavior to an
individual’s work-related performance.
With the above in mind, I performed an exploratory study of Icon Medialab
(Icon), a rapidly growing firm within the new industry of internet consulting. Founded
in 1996, the company had grown to 240 employees with offices in eight countries
within two and a half years.66 In addition, Icon was considered to be on the ”bleeding
edge” of knowledge-intensive companies typical of the new economy. Icon
specialized in technologically complex digital communications solutions for large
multinationals as well as for start-ups with radically new business models. Icon’s
products include business-to-business, business-to-consumer, and consumer-to-
business internet-based solutions. A major objective for Icon management was to
ensure not only the development and use of the latest internet technology, but also the
reuse of this technology in subsequent projects. However, this is a difficult challenge
since the pace of technological development is so rapid with products often becoming
outdated within six months or less from development. A final reason for choosing
Icon was that its employees in all functions were not only extremely adept at using
new internet-based communication media such as bulletin boards, chatrooms, email,
etc. but they also used these to a high degree in their everyday work.67
66 At the end of 1999, Icon Medialab had 1056 employees spread across the globe in 19 offices in 13 countries.
67 One potential explanation is that the average age at Icon was 29.9 years.
322 ARTICLE THREE
My first step was to conduct a substantial number of interviews of various

functions and levels at Icon. I then administered a detailed questionnaire to every
employee in the firm aimed at developing an understanding of the sources of
knowledge that each individual used in the course of his or her everyday work. With
this data, I was then able to build a rich picture of an individual’s knowledge networks
inside the firm, and more interestingly of an individual’s networks that reached across
the boundaries of the firm. I then linked these knowledge flow patterns to individual
performance on various dimensions in order to provide clear evidence regarding the
value of these knowledge flows.
This article is organized as follows. In the following section, the Communities
of Practice literature is reviewed briefly. This literature provides the foundation for
the conceptual model, and the specification of six hypotheses linking knowledge flows
to individual performance. Section three describes the research methodology and
provides a brief description of Icon Medialab. Section four reports on the results of
the empirical study while the last section provides a discussion of the results and the
implications of the research for theory and practice.
BACKGROUND
In today’s highly competitive environment, traditional industries are merging at the
same time as completely new industries are emerging. With these changes comes the
birth of numerous new firms that is considered important to the vitality of these
industries. In addition, many of these new firms experience a period of rapid growth
as they struggle to gain leading market positions (Aldrich & Fiol 1994, Davidsson et
al. 1996, Delmar & Davidsson 1998a, 1998b). However, the ability to succeed is
becoming increasingly difficult due to shrinking product lifecycles, the need for
integration across an increasing diversity of technologies in products and services, and
increasing levels of competition from new competitors crossing not only geographical
but industrial borders as well (Boland & Tenkasi 1995, Purser, Pasmore, & Tenkasi
1992). All of this puts increasing pressure on these high-growth firms to do a better
job of gaining access to new knowledge in their business environments while at the
same time leveraging their existing knowledge across the firm (Bartlett & Ghoshal
1989, Doz & Hamel 1997, Drucker 1990, Hedlund & Nonaka 1993). While there has
been considerable interest in studying the creation and leverage of knowledge at the
firm level (see for example, Spender 1989, Nonaka 1991, 1994), until recently there
has been little interest in the individual and the manner in which his or her knowledge
contributes to the knowledge of a firm.
One area that looks more closely at the individual is the knowledge-based view
of the firm. This view argues that a firm should be understood as a social community
of individuals who have a shared identity (Kogut & Zander 1992, 1995). As
individuals work together over time, they develop shared mental models, a common
language, and common behaviors. This shared identity lowers the costs of
communication between the firm’s members and results in explicit and tacit rules of
coordination as routines are built over time. In addition, a common language enables
members to codify their tacit knowledge. Relative to individuals outside the firm,
employees can then easily access and reuse this codified knowledge as they share the
same communication code and mental models as those who codified it (Nonaka &
Takeuchi 1995). In this manner, it is relatively easy for employees to search the
company for advice or existing solutions (Constant, Sproull, & Kiesler, 1996). One
conceptual lens through which the firm as a community of individuals can be studied
is the emerging community of practice body of research, which is the subject of the
next section.
In an observed that there was a variance between the organization’s formal description
of work and the way in which the actual work was performed. When the technicians
were faced with problems for which the formal structure often did not provide
solutions, they relied on the organization’s informal systems for help, such as
storytelling, conversation, mentoring, and experiential learning. (Brown & Duguid
1991, Orr 1990, Snyder 1997, Wenger 1997). Individuals collaborated with each other
through an emergent and fluid structure of relationships and engaged in patterns of
exchange and communication to reduce the uncertainty of their tasks (Pava 1983,
Purser et al. 1992). Thus, the procedures required to fulfill the tasks were developed
informally as the workers performed their tasks, demanding the creation and use of
knowledge along the way (Purser et al. 1992, Stebbins & Shani 1995). These
informally established groups of collaborating individuals were then named
communities of practice.
Communities of practice have no ethnographic study of Xerox service
technicians in the late 1980s, it was real boundaries and are in a constant state of
evolution as members come and go and commitment levels fluctuate. This fluidity
creates difficulties when management wants to pin down communities of practice,
determine their boundaries, and develop some form of recipe to manage them. Indeed,
it is argued that this is not possible due to the pure informal nature of communities of
practice (Wenger 1998). Thus, we must satisfy ourselves at this point with a definition
that captures this fluidity and intangibility.
A group of people informally and contextually bound who are applying a

common competence in the pursuit of a common enterprise (Brown &
Duguid 1991, Lave & Wenger 1991, Snyder 1997, Wenger 1998)
Operationalizing Communities of Practice

While the amount of papers and articles focusing on communities of practice continues
to grow, few researchers have attempted to understand the relation between
communities of practice and performance. This is understandable because they are –
by definition extremely hard to pin down. Any individual can potentially be involved
in numerous communities of practice, varying from one’s immediate workgroup to a
set of internet contacts on the other side of the world. Moreover, the process of
324 ARTICLE THREE
defining the membership of communities of practice apparently takes away their very
essence, because they thrive on their informal nature. Bearing this in mind it is not
surprising that the community of practice literature is populated with ethnographies
and case studies rather than surveys or experiments.
The approach in this paper is to bring the community of practice thinking down
to the level of the individual. Rather than attempt to define the community of practices
within and across the firm’s boundaries, it is assumed that an individual’s performance
at work is associated with the extent to which he or she is a member of various
communities of practice, including those facilitated by electronic means. Thus, by
measuring the patterns of communication of the individual with various groups of
people, and through various different forums, one can predict to some degree his or her
performance. In the next section, this idea is developed into a series of testable
propositions.
CONCEPTUAL FRAMEWORK AND PROPOSITIONS

As discussed earlier, competitive advantage is built on the firm’s ability to acquire new
knowledge from outside the boundaries of the firm while at the same time leveraging
the existing knowledge within the firm. However, this leaves us with a vague idea
regarding on which level knowledge acquisition and leverage occur because they can
potentially occur at all levels – the individual, the group, the business unit, and the
firm. As suggested by Hedlund (1994), it is the ability to transfer knowledge between
levels of analysis (e.g., from the individual-level to the firm level and vice versa) that
is valuable, and indeed one of the major characteristics that makes the firm unique.
In this paper, two levels of analysis are of interest: the individual and the
community of practice. The logic here is that individuals are able to draw from their
communities of practice to solve problems they encounter in the course of their work,
and that they also contribute back to these communities in a reciprocal manner. Thus,
the extent to which an individual is actively involved in communities of practice will
ceteris paribus be associated with superior performance at work.
But as previously noted, there are significant methodological problems in
studying communities of practice. The primary concern is that the concept is typically
defined in such a way that all informal interactions, inside or outside the firm, could
represent participation in communities of practice. If this broad definition is accepted,
then the concept becomes very difficult to research in a rigorous manner because
nothing is excluded. The theory, in other words, cannot be falsified.
The approach taken here, as hinted above, is to move down to the level of the
individual and to then examine the way in which that individual acquires new
knowledge to address work-related problems. Some knowledge, as we will see, is
gained through access to “codified” sources such as the internet or company databases,
but most is gained through interaction with other people in the firm and outside. The
premise, in other words, is that the frequency and quality of the interaction an
individual has with specific groups of individuals is a manifestation of the
communities of practice with which he or she is involved, and that such interactions
will have a positive impact on his or her individual-level performance.
The conceptual framework in figure 1 illustrates this approach. Individual level
performance, I argue, is a function of the various ways knowledge is acquired by the
individual, and the sources of that knowledge can be divided into (1) internal vs.
external sources, and (2) tacit vs. codified sources. In addition there are many other
factors contributing to individual level performance, some of which are empirically
examined as controls.
Figure 1 Conceptual Framework
Internal Sources
•CP Interaction
•CP Socialization
•Codified Sources
External Sources Individual Performance

•CP Interaction •Creativity
•Codified Sources •On-time delivery
Control
•Education
•Time at Current
Employer
•Work Experience
• Openness
Proposition Development
The first proposition follows directly from the discussion about the nature of
communities of practice inside the firm. As stated above, individuals within
organizations are thought to be members of numerous communities of practice.
Informally collaborating within these communities, individuals create and exchange
tacit knowledge in a more effective means than through formal structures and systems
(Schön 1983, Snyder 1996, 1997). Being an active member of communities within the
organization thus implies a high degree of collaboration and interaction with other
members through primarily face-to-face but also non-face-to-face interactions. A high
degree of interactions with other community members should therefore lead to a
greater individual development of task-related knowledge and thus higher
performance. Thus we have our first proposition.
PROPOSITION 1: The greater the level of individual personal interaction

with members of communities within the firm, the higher the level of
individual performance (creativity, on-time).
326 ARTICLE THREE
For an individual to truly become a member of a community of practice and

access the community’s knowledge, it is argued that he or she must not only have a
high degree of interaction, but also become an “insider” through the development of
shared trust with other members of the community (Lave & Wenger 1991, Snyder
1997). Through a high level of shared trust, the member learns of other members’
mistakes and breakthroughs through storytelling and narration of work-related
happenings. While trust is difficult to measure, and particularly so when the
community in question is not clearly specified, one manifestation of it is in the
existence of social contacts outside of work. Thus:
PROPOSITION 2: The greater the level of social interaction with community

members outside of work, the higher the level of individual performance
(creativity, on-time performance).
The two propositions above are concerned with an individual’s participation in

communities of practice within the boundaries of the organization. However, it is
central to the concept of communities of practice that they also spread across
organizational boundaries, through professional or technical relationships
(Venkatraman & Henderson 1998). These communities may involve members from
the suppliers, customers, or even friends working on similar tasks in other companies.
A high degree of interaction with members of communities that cross organizational
boundaries can be expected to broaden the individual’s knowledge through the
exchange of knowledge from outside the firm. Thus, much in the same manner as
proposition 1, proposition 3 becomes the following:
PROPOSITION 3: The greater the level of individual personal interaction

with members of communities that spread across organizational
boundaries, the higher the level of performance (creativity, on-time).
Propositions four and five are concerned with the acquisition of knowledge
through codified sources. The spread of the internet and the development of intranets
are factors that have led to this explosion of rapidly accessible codified knowledge.
While face-to-face collaboration with a community of practice is understood to be the
primary channel for the development and exchange of primarily tacit knowledge, this
channel can also be supplemented with non-face-to-face lateral written
communications taking the form of electronic communities. Many organizations are
in the process of implementing electronic communities to promote knowledge sharing
between organizational individuals (Alavi & Leidner 1999, Davenport & Prusak 1998,
Fulk & DeSanctis 1995). In addition, the use of codified sources of data such as
company documents facilitates an individual when solving work-related tasks. For
example, access to an internal document can help an individual to avoid reinventing
the wheel, thus facilitating the completion of a work-related task. This codification
and documentation of knowledge within the firm is one of the main thrusts of
management in organizations in order to ensure the transfer and application of

knowledge throughout the firm. Thus, we have our fourth proposition.
PROPOSITION 4: The greater the use of internal codified sources of

information, the higher the level of individual performance (creativity, on-
time).
In addition to company-specific codified knowledge, individuals also have access

to numerous sources of codified knowledge outside the firm. In today’s fast-changing
world, the knowledge required to solve a new, challenging task may not exist inside
the firm, and thus the individual may have to search outside the firm for help. An
individual can participate in numerous electronic communities that extend across
organizational boundaries (Hagel & Armstrong 1997). Thus, an individual can
communicate with thousands of others anywhere across the globe, regardless of
demographic characteristics, organizational setting, or local culture (Hinds & Kiesler
1995, Sproull & Faraj 1995, Faraj & McClure Wasko 1999). We then have our fifth
proposition.
PROPOSITION 5: The greater the use of external codified sources of

information, the higher the level of performance (creativity, on-time).
Finally, we have the rather general proposition that the extent to which the above
approaches to knowledge acquisition affect performance will be contingent on the
nature of the task being performed. Space limitations prevent a detailed discussion,
but one would expect ceteris paribus that the less routine, the more intellectually
challenging, and the more fast-changing the work, the more important it would be to
have ready access to personal and codified sources of knowledge. In terms of the
specifics of this study, those individuals who work in software programming and web
design have been separated from those doing other tasks (such as administration, sales,
and management), on the basis that the former group are likely to rely more on
knowledge acquisition from a variety of sources to undertake their work effectively.
Thus:
PROPOSITION 6: The relationships put forward in propositions 1-5 will be

stronger for “technical” employees (software programmers, web designers)
than for other employees.
Many other factors are also expected to be associated with individual

performance. In this study, the education level of the individual, the amount of time
he or she has spent with Icon Medialab, and their general work experience are also
measured. In addition their perceptions of how “open” the work environment is at
Icon are measured because it is a factor that is likely to effect their propensity to
exchange information with others.
328 ARTICLE THREE
Finally, it is worth observing here that individual-level performance is not a uni-

dimensional construct. At the very least, we would expect to see a split between
“exploration” and “exploitation” (March 1991) where exploration would be
manifested as creativity or the development of novel solutions, while exploitation
would be manifested in the ability to get work done on time and on budget. However,
given the exploratory nature of this research I have not specified any a priori
expectations regarding the type of performance associations we expect to see.
METHODS
Sample and Analysis

The research was undertaken in a single firm, Icon Medialab. While the objective in
the future is to broaden the investigation to other firms, it makes sense to begin in a
single case and then to re-evaluate on the basis of the findings from that study. As
already mentioned, the choice of Icon was motivated primarily on the basis that it is a
high growth, quintessential “IT-intensive” firm, in which a large proportion of the
employees are working on a day-to-day basis with the latest internet technology.
Many of these employees, it turns out, are interacting frequently with “communities”
of “techies” whom they have never met. As such, this setting represents a fascinating
test of the communities of practice concept.
It should be noted that Icon is based in Stockholm. This was not only
convenient, but Stockholm is also quite an opportune location for studying such a firm
because Sweden is at the forefront of digital communications technology. The country
has one of the highest penetration rates in the world of mobile telephones and internet
subscriptions per capita, and Stockholm is a recognized high-technology “cluster”.
Icon is one of many recent start-up Internet firms in the area (founded 1996), and one
of the world’s best 350 small companies according to Forbes (Forbes 1998). A
description of the company follows.
Two phases of data collection and analysis were conducted. The first phase
was conducted at the Swedish office, in which thirty in-depth field interviews were
held from May 1998 to June 1998. People at different areas of the company, e.g.,
corporate management, business development, sales, and different production
competencies, were interviewed for one-and-a-half to two hours each. Interviews with
management were conducted first in order to understand the formal structures that had
been put in place to facilitate knowledge acquisition mechanisms. Extensive written
material was also collected from the companies.
The second phase of the data collection during the fall of 1998 involved a
questionnaire sent to all 242 employees of Icon Medialab at their local offices.
Questionnaires were then sealed in individual envelopes and returned to us either by
mail or directly. Of the 242 questionnaires, 203 usable questionnaires were collected,
an 84% response rate. Throughout the data collection process, individuals were
assured that their responses would be kept confidential and that all results would be
presented on an aggregated level. In addition to these individual questionnaires, each

of the managing directors of the eight subsidiaries and seven managers at the
Stockholm office were asked to complete a questionnaire relating to the performance
of the individuals at their office. The average age of the respondents was 29.9 years
with an average of 385 days employed at Icon and 4.8 years experience in their
competence. The sample was 30% women.
In terms of the split between functions, there were 72 respondents working in
technically oriented functions (e.g. programmers, web-page designers) and 131
working in the non-technically oriented functions (e.g. sales, administration). As
discussed in the previous section, this split is important in terms of understanding the
types of communities of practice individuals are likely to develop.
Measures
Dependent Variables
Several different approaches exist for measuring performance, including both
subjective and objective measurements. For the purposes of this study we used two
different subjective dependent variables that measure individual performance,
creativity and on-time performance. As discussed above, these measures represent the
two dimensions of performance of exploration and exploitation where exploitation is
manifested as creativity or the development of novel solutions and exploitation
manifested as the ability to get work done on time and on budget (March 1991).
While it is somewhat difficult to distinguish between these two measures, we do feel
that it is important to measure both since it is often difficult to develop solutions that
are highly creative but that are also on budget and on-time.
1) Creativity – Individuals were asked to answer 3 questions that created a creativity
scale (Sjöberg & Lind 1994). These were based on a seven-point scale from 1,
“strongly disagree”, to 7, “strongly agree” (three items, α = .64).
2) On-time performance – The final performance measure asked respondents to
answer to what degree they felt they delivered their work on-time on a seven-point
scale from 1, “strongly disagree”, to 7, “strongly agree” (two items, α = .66).
In addition, we asked the managers in each of the offices to rate the
performance of each individual reporting to him or her on two different items: ability
to meet superior’s objectives and to develop creative solutions. While the two items
were strongly correlated with each other (r = .75), the correlation with the various self-
reported performance measures was very weak (between 0.05 and 0.28). After
discussing this matter with several of these individuals, it became clear that the
managers often had remarkably limited contact with many of their direct reports, and
that they could not easily assess their performance. We therefore concluded that the
330 ARTICLE THREE
self-reported performance measures were more valid, an observation that is consistent

with a number of previous studies (e.g., Heneman 1974; Wexley et al. 1980)68.
Independent Variables
These variables included the different dimensions of the knowledge acquisition
processes and we have chosen to split them on the external vs. internal dimension.
The external mechanisms consisted of two measures. The first measure which relates
to proposition 5, Codified - external, asked respondents to answer on a seven-point
scale the frequency of use of external knowledge sources. These sources included
traditional sources such as externally produced books or journals in addition to
recently developed sources such as Internet web pages or Internet discussion forums
(five items, α = .73). The second measure, External community interaction, was
measured on a four-point scale relating to the degree of interaction on work-related
matters with customers and friends. Respondents were asked how often they initiated
the interaction as well as how often the external party initiated the interaction (four
items, α = .80) and relates to proposition 3.
The second group of variables, internal mechanisms, consisted of 7 different
measures. The first measure relating to proposition 4, Codified – internal, asked
respondents to answer on a seven-point scale the frequency of use of internal
knowledge sources. These included using the company’s intranet as well as materials
such as documents that were produced internally by Icon (three items, α = .64). The
next measure, Interaction with internal community, that relates to proposition 1 was
based on a four-point frequency scale on two dimensions. The first one was based on
whether it was the respondent who initiated the interaction and the second one based
on whether the other party initiated the interaction. This measure was built upon the
interaction with others within the same function, others within one’s workgroup, and
others outside of one’s workgroup (three items, α = .64). For example, the workgroup
of a programmer included art directors and web designers as these three functions
comprised the production team for each project. Those outside of the programmer’s
workgroup included those in support functions, e.g., sales, or those in management
functions, e.g., human resources. The final measure, Social contact, was measured
through the level of social contact outside of work with any individual throughout the
organization. This was measured on a simple 1-2 scale, 1 for no and 2 for yes (3
items, α .86) and relates to proposition 2.
Control Variables
These variables included level of education (1-5 scale), time employed at Icon (no. of
days), related work experience (years)69, and openness. Openness was created to
measure the level of openness at Icon perceived by the individual on a seven-point
68 It is worth noting in passing that the significant correlates with manager-rated performance were (a) age of
employee, and (b) lack of socialization with other people outside of work. In other words, managers believe that
older employees without social contacts with colleagues are the better performers!
69 In order to avoid multicollinearity problems, we decided not to include age as it correlated highly with related
work experience.
scale from 1, “strongly disagree”, to 7, “strongly agree” (11 items, α = .65). Summary
statistics for the control variables and the other variables are presented in the table in
the appendix.
The propositions were tested through a series of stepwise regression models.
The stepwise approach was chosen primarily because of the small sample size and the
relatively large number of independent variables. Also, the exploratory nature of the
study makes it appropriate to work with a rather larger number of independent
variables than would normally be the case. The plan in future iterations of this
research will be to move towards a more carefully specified model.
Company Description
Icon Medialab was founded in March 1996 in response to the rapid growth of the
internet 70. The company’s mission was to facilitate the creation of competitive
advantage for its customers through the incorporation of the internet in customer
operations. Products and services included internet homepages, intranets, extranets,
and e-commerce solutions. Icon Medialab’s clients ranged from the Swedish Postal
Service and Compaq to British Petroleum and Volkswagen. The company posted
sales of SEK 65 million for the fiscal year ending April 1998 (SEK 13 million in 1997)
and at the time of this study had 242 employees with 46% of these in Sweden. The
remaining employees were spread throughout offices of 10-25 employees in Spain,
USA, Finland, Denmark, Germany, Belgium, and the UK, with new offices planned
for France and Norway. Table 1 provides some key figures for Icon Medialab.
Table 1 Figures for Icon Medialab

1999 1998 1997 1996 (10 mos.)
Sales (SEK mln) 412 132 43 4.59
Number of Employees 1056 300 141 62
Number of Offices 19 9 6 3
Number of Countries 13 9 6 3
A strategy of rapid global growth was developed by the founders at the

company’s inception. One of the means by which Icon hoped to achieve profitable
growth was through the reuse of knowledge developed throughout its different
projects. In fact, management set a target that more than 50% of all projects should
include already proven successful products or services. Thus, Icon Medialab invested
heavily in building its structural capital, with the key objectives to transfer and reuse
knowledge complemented with follow-up and reporting.
In addition, Icon Medialab was unique in its representation of a mixture of
competencies under the same organizational umbrella. These disciplines included
Technology, Design, Usability Engineering, Statistics and Analysis, Media and
70 The digital communication market is among the fastest growing markets ever. In a report by the International
Data Corporation, the market for internet services is predicted to grow from USD 2.5 billion in 1996 to USD
13.8 billion in 2000.
332 ARTICLE THREE
Entertainment, and Business Strategy, representing the 6 sides of the “Icon Cube”.
Thus, Icon Medialab brought together art directors, behavioral scientists, copywriters,
journalists, scriptwriters, animators, TV-producers, software programmers,
management consultants and web designers, with accounting, personnel, and
administration completing the organization.
RESULTS
Descriptive Statistics
Before moving to the results of the regression analysis, it is important to describe the
patterns of knowledge acquisition among the Icon employees, and in particular
whether there are any significant differences between technical employees and others
(cf. proposition 6). Table 2 presents a comparison of the means of the use of different
sources of knowledge for the two groups at Icon. While there is no significant
difference in the level of internal community interaction and social contact outside of
work, Codified – internal sources, Codified - external sources, and Interaction with the
external community do differ significantly. The technically oriented people tended to
use external codified sources of information more than the non-technically oriented
people while the opposite is true for the internal codified sources. Based on our scale,
technically oriented people used external codified sources once a week on average,
while the non-technically oriented people used these sources between once to twice a
month. With regard to internal codified sources, non-technically oriented people
accessed these about once a week and the technically oriented people closer to once a
month. In addition, the level of interaction with the external community was higher
for non-technically oriented people than for those who are technically oriented. Non-
technically oriented people interacted with the external community an average of
about two times a week while the non-technical people interacted about two times a
month.
Table 2 Comparison of Means

Variable Nontech. Tech. t
External
1. Codified - Externala 3.43 4.00 3.30**
2. Interaction with External Communityb 2.96 2.46 -4.00***
Internal
3. Codified – Internala 3.95 3.20 -3.90***
4. Interaction with Internal Communityb 3.07 3.15 -0.88
5. Social Contact Outside Workc 1.36 1.34 -0.35
a
1 – yearly or less, 2 – four times a year, 3 – once a month, 4 – once a week, 5 – twice a week, 6 –
once a day, 7 – several times a day
b
1 – rarely or never, 2 – once a month, 3 – twice a week, 4 – a few times a day
c
1 – no, 2 - yes
** p < .01
*** p < .001
Propositions
Individual performance was measured using two different constructs, Creativity and
On-time performance. In addition, we have reported two different models for each
dependent variable– one for the whole sample, and one for just the technical
employees. The purpose of this split is to see if the results differ significantly when
one just considers technical employees.
Self-reported Creativity – Whole Sample

In table 3, limited support for propositions 5 and 2 is received for the whole sample
size. Consistent with proposition 5, the coefficient of the use of External codified
sources is positive (p < .10). In addition, the coefficient of Social contact outside of
work is positive (p < .10) and is consistent with proposition 2. Thus, the higher the
level of social contact and the higher the use of external codified sources, the higher
the level of individual creative performance. In addition, the control variables were
also highly significant: Education (p < .05), Office openness (p < .01), and Related
work experience (p < .001).
Self-reported Creativity – Technically Oriented

Table 3 also provides the results for only the technically oriented people. Stronger
support is provided for proposition 5 since the coefficient for Codified external is more
significant (p < .05) than in the above regression. Social contact is the same as above
(p < .10), thus providing limited support for proposition 2. In addition, there was only
one significant control variable, Openness (p < .05). Of interest is that the adjusted R2
is the highest for this regression than any of the others, 0.45.
Self-reported On-time Performance – Whole Sample

Support for proposition 4 is given in table 4 as the coefficient is positively related (p <
.05). Thus, on-time performance is positively related to the use of codified internal
sources. In addition, two control variables turned out negatively related to
performance: Education (p < .05) and Time at Icon (p < .10). This regression had the
weakest R2 of the regressions, 0.11.
Self-reported On-time Performance – Technically Oriented.

Table 4 shows that there is support for proposition 4 as the coefficient, Codified
internal, is positively related to On-time performance (p < .01). In addition,
proposition 5 was rejected as the coefficient, Codified external, was negatively related
to performance (p < .05). Only one control appeared significant, Openness (p < .10),
in this equation.
334 ARTICLE THREE
Table 3 Results of Regression Analysis of the Relationship between Knowledge

Acquisition Mechanisms and Self-reported Creativity
Variable Model 1 Whole Sample Model 2 Tech. Only
Prop. Var.Incl. Var.Excl. Var.Incl. Var.Excl.
Control
1. Education .18* -.00
2. Work Experience .31*** .10
3. Time at Icon .03 -.22
4. Office Openness .30** .42*
External Knowledge
=
5. Codified - External 5 .15 .35*
6. External Community 3 .08 -.22
Interaction
Internal Knowledge
7. Codified - Internal 4 .08 .11
8. Internal Community 1 .07 .06
Interaction
9. Social Contact Outside 2 .17= .30=
Work
R2 .28 .51
∆R2 .25 .45
F for ∆R2 7.97*** 8.31**
=
p < .10 * p < .05 ** p < .01 *** p < .001
Table 4 Results of Regression Analysis of the Relationship between Knowledge

Acquisition Mechanisms and Self-reported On-time Performance
Variable Model 1Whole Sample Model 2 Tech. Only
Prop. Var.Incl Var.Excl Var.Incl Var.Excl
Control
1. Education -.23* -.11
2. Work Experience .04 -.04
=
3. Time at Icon -16 -.09
=
4. Office Openness .08 .28
External Knowledge
5. Codified - External 5 -.04 -.31*
6. External Community 3 -.10 -.06
Interaction
Internal Knowledge
7. Codified - Internal 4 .23* .55**
8. Internal Community 1 .11 .26
Interaction
9. Social Contact Outside Work 2 -.11 .08
R2 .13 .48
∆R2 .11 .42
F for ∆R2 5.25** 7.43**
=
p < .10 * p < .05 ** p < .01 *** p < .001

Table 5 provides an overview of the support for the different propositions from the
different regression models. Altogether we see support for propositions 2, 4, and 5.
Moreover, what is interesting here is that we see such different results for the two
dependent variables. Again, as discussed above the study looked at two dimensions of
performance, creativity and on-time performance that represent exploration and
exploitation, where exploitation is manifested as creativity or the development of
novel solutions and exploitation manifested as the ability to get work done on time and
on budget (March 1991). Following is a discussion of the regression findings related
to the qualitative findings.
Table 5 Support for Propositions

Creativity On-time
Proposition Whole Tech. Whole Tech.
1. Interaction with Internal
Community
2. Social Contact outside Work .17= .30=
3. Interaction with External
Community
4. Codified - Internal .23* .55**
5. Codified - External .15= .35* -.31*
=
p < .10 * p < .05 ** p < .01 *** p < .001
Creativity
Taking creativity first, we see social contact outside of work and the use of external
codified sources of information (internet communities and the like) as the significant
predictors. Building on our qualitative findings, the impression one gets is that
technical employees attach great importance to their external internet-based
relationships as sources of ideas and as ways of solving tricky problems. Several
programmers even stated that they preferred to go first to their internet community or
use their private email list for help instead of asking someone at their own company
even if he or she was sitting at the next desk. Through interviews, several reasons
were found. The first was that by posting a question in an open forum on the internet,
people were not obligated to help. Instead those who wanted to help could do so in a
voluntary fashion. By reaching out to the electronic community for help, one did not
disturb a colleague at work who had his or her own schedule and deadlines to meet.
Another reason found was that people could access a much broader source of expertise
than at their own company. In many instances, individuals claimed that there was no
“critical mass” internally, especially when discussing the intranet, within Icon.
However, this critical mass could be found on external websites and communities.
Members of external communities worked at different types of companies all over the
world, yet they worked on the same type of problem. Thus, it was felt that this
336 ARTICLE THREE
enabled one to gain access to the latest thinking within one’s field, especially since the
change of pace within this industry was so rapid.
To turn the discussion to the second predictor, Social contact, individuals
became members of a tightly knit community of practice through extensive social
contact outside of work. During this social contact, these individuals discussed the
difficult problems encountered during the day, the responses received from the
electronic community, and how they then attempted to solve the problem. The latest
solutions or tips from both the outside communities and one’s own work were passed
between the members of the community. In this manner, these community members
socially constructed their world through the narration of stories, turning incoherent
data into coherent information. This enabled them to gain insights into the work they
were performing, allowing them to be more creative in their daily work. What is
interesting here then is that it is the combination of the use of an external community
with one’s internal community. As ideas cross community boundaries, resulting in the
cross-fertilization of communities, knowledge is combined to foster creativity. Based
on these findings and previous research, we then developed the term Netovation to
describe this creative performance that was fostered through the use of the internet
(Teigland 1999).
On-time Performance
In terms of achieving on-time performance, a very different picture emerges. Here, the
use of internal codified sources of information is a positive predictor of on-time
performance, while the use of external codified sources is a negative predictor. This is
entirely in keeping with intuitive expectations. Building relationships with external
communities and creating unique or “elegant” solutions on the basis of those
relationships works well when creativity is the objective, but it is a strong negative
when on-time delivery matters. Gathering information from the outside takes time
because first either the sources must be located or one must wait for someone to
voluntarily help. And once the information or help is received, it must be assimilated
into the context of both the problem and the company’s way of doing things. This
may take considerable time depending on the complexity of the information and the
problem.
In addition, reciprocity within these electronic communities is necessary in
order to become a true member. In other words, to be able to ask the other internet
community members for help, one must prove that one also gives back to the
community through providing help to others when asked. This returning of help then
results in the individual performing work for others outside the company. This then
takes away time from the individual’s internal responsibilities, potentially leading then
to poor on-time performance.
Thus, on-time performance can best be achieved by re-using existing solutions
that can be accessed through the firm’s intranet or company documents. To give an
example, Icon’s intranet included a programming module database that included both a
description for the sales force and a technical description for the programmers.
Programming modules were building blocks of programming code that could be

reused in a number of customer projects, such as a discussion forum, telephone book,
or conference room booking system. In addition to a technical description, the module
list also specified how many hours were required to develop the module. This
information was added to help determine the pricing and planning of future projects.
Participation in Communities of Practice

While it was found that both internal and external sources of codified information
were significant for different types of performance, it was quite puzzling that no
support was found for the relation between performance and the frequency of
interaction with members of communities of practice either internally or externally
located. As we saw above (table 3), both non-techies and techies used codified
internal and codified external sources of information to a higher degree than
interaction with communities either internally or externally. This seems to go against
the community of practice literature that says that individuals draw upon their
communities to help them in their everyday work. However, a deeper analysis of
community interaction is necessary. In addition to asking the set of questions
regarding the frequency of use of information sources, a set of questions was asked
which looked at the helpfulness of information sources when solving a particularly
difficult problem. This was designed in order to get a picture of the usefulness of
different information sources as opposed to the frequency of use. The results are
presented in table 6 below.
As we can see, non-techies ranked informal discussions with someone in their
office (not their immediate superior) as the most helpful source while techies ranked
these informal discussions as the second most helpful. Thus, we see that internal
communities are of importance when solving a particularly difficult or challenging
problem. This is in line with the communities of practice literature that says that
individuals do turn to their community for help with tasks that fall outside of the
routine way of doing things. What is of interest here is that techies ranked internet
web pages as the most helpful information sources over interaction with communities.
Some individuals even ranked internet discussion forums and electronic communities
as the most helpful. As discussed above, one of the reasons was that the technology
was changing so rapidly thus in order to access the latest thinking within the field, it
was quicker to go to external sources through the internet to get an idea of what the
answer might be and to then discuss this with internal community members to adapt it
to the specific task at hand.
338 ARTICLE THREE
Table 6 Sources of Information Used for Solving a Difficult Problem

Think back to the last really difficult Whole
problem that you solved. Rank the Sample Nontech. Tech.
following sources of information (n= 200) (n=131) (n=69) t
Scanning or reading externally printed
3.15 3.04 3.33 1.05
materials
Internet web pages (company pages,
3.39 2.99 4.13 4.14***
FAQs, help desks etc)
Internet discussion forum, electronic
1.89 1.64 2.35 3.26**
community
Private email list
2.04 2.00 2.10 0.43
Icon’s Intranet
1.72 1.76 1.64 -0.67
Informal discussions with project
3.01 3.20 2.64 -1.83=
manager, competence coach, leader
Informal discussions with someone
4.04 4.11 3.90 -0.74
other than above in your office
Informal discussions with someone in
2.31 2.52 1.91 -2.46*
another Icon office
Informal discussions with customers
2.15 2.35 1.75 -2.52*
Informal discussions with someone at
1.79 1.76 1.84 0.42
another Internet company
Scheduled meetings
2.50 2.73 2.04 -2.70**
=
p < .10 * p < .05 ** p < .01 *** p < .001
The cells indicate the average rating for the item in question. Respondents were asked to indicate up to
five sources of information, where a score of 6 = most helpful, 5 = second most helpful, 4 = third most
helpful, 3 = fourth most helpful, 2 = fifth most helpful. A score of 1 meant that the source of
information was not mentioned. Thus, a higher average rating means the source was more helpful in
solving problems.
However, another quite interesting reason for the significant use of internet
webpages and electronic communities was uncovered during the qualitative phase –
that of prestige. Several interviewees commented that some individuals feared making
mistakes or making themselves look stupid by asking others at Icon for help. So, they
turned to the internet where “no one knows if you’re a monkey”. Another aspect was
that it was seen as prestigious if one belonged to some of the closed internet
communities. Some of these qualitative findings seem to be in accordance with Zipf’s
Law of Least Effort (1949), which argues that individuals when choosing a path
towards a goal are more interested in minimizing effort than maximizing gain. This
effort includes both physical as well as psychological effort. Thus, although asking
another programmer within the company for help may result in a larger gain, it may
cost the knowledge seeker more in terms of psychological effort, showing that he or
she does not know the answer71. The following quotation exemplifies this,
“Sometimes it seems that there are some programmers who are afraid to ask for help.
So, instead they just sit there and work for hours trying to solve something instead of
asking.” In relation to the community of practice literature, this psychological effort
may be higher for those individuals who are not legitimate members of a community.
These individuals do not have a feeling of shared trust with the core members of a
community and experience a higher level of psychological effort by asking questions
to the internal communities.
Turning to Icon, at the time of data collection, the company was in a very high
rate of growth both in terms of the number of employees as well as in the number of
offices. The Stockholm office was only two and a half years old with four of the
offices less than one year old and an average individual tenure at the company of 385
days. Thus, many of the individuals had not yet had time to become legitimate
community members. In addition, in many offices other than the Stockholm office
there was only a handful of individuals within each function, and potentially even
fewer for the technical functions. Thus, the ability to interact with internal community
members was limited in many respects by either lack of members or the insufficient
tenure to become a community member. The importance of becoming a member
within the community in order to access help was exemplified in the following
citation. “Other programmers call me up pretty often because they have heard about
me. But it feels really strange when you don’t know them. It then becomes a matter
of prestige - why should I help you?”
Implications for Theory and Practice

This study raises a number of very interesting theoretical and practical issues. First,
what do we make of the concept of the community of practice in the light of our
findings? Some support was clearly found, in terms of the importance of work-related
and social interaction with other community members. But there were also some
surprises, notably the importance of so-called “internet communities” as sources of
knowledge for technical employees. This is a curious discovery because these
“communities” exhibit many of the characteristics of communities of practice –
reciprocity, identity, and so on—but the individuals involved have typically never met,
and they work through what is by definition a codified exchange of information, which
goes against other aspects of the theory. One programmer commented,
I’ve been really active in the internet community for a long time. I’m in
contact with a group of about 20 people who are experts at what they do.
But I have never met them physically. But it doesn’t matter because on the
internet we have always been friends. It’s just like when you used to go
71 This also has some parallels to the work done by Edmondson (1999) on team psychological safety. Team
psychological safety is defined as the shared belief by the team that the team is safe for interpersonal risk-taking,
thus encouraging people to express their ideas without fearing that they will be rejected. This belief stems from
mutual respect and trust among the team’s members.
340 ARTICLE THREE
snowboarding 10 years ago. If you were somewhere and saw another

snowboarder, you said hi and then you’d hang out together in the evening.
Just because we snowboarded and there were so few who did it. We were
on the same level…we knew where we had one another.
The development of these electronic communities has added a spoke of a new

dimension to the community of practice literature. Whether or not these electronic
communities can be considered communities of practice is an issue currently being
debated among scholars. According to Lave & Wenger (1991), socialization (i.e.,
face-to-face contact) among members is an important factor in the building of a
community of practice. Nonaka & Takeuchi (1995) further explain that socialization
requires the sharing of tacit knowledge that in turn requires a shared space among the
community of practice members. Since communication through the internet is textual
among individuals who often are spread across the globe, it is argued that participating
individuals do not have a shared space, thus there is no socialization, and more
importantly, no sharing of tacit knowledge. Followers of this view would then argue
that these internet-based communities are not communities of practice in the “true”
meaning of the term. As a result, we need to revisit certain aspects of the community
of practice to better understand which theories are still valid and which need to be
adapted to the new empirical contexts.
Secondly, our results provide some preliminary support for the idea that the
firm is a vehicle that brings together members from different communities of practice
that are more professionally oriented. Thus, a programmer may belong to a C++
community that spreads across many firms’ boundaries while the other functions at
Icon, e.g., management consultant, art director, etc., do the same. Icon then brings
together these different communities by hiring members and incorporating them into
project teams. Professional knowledge flows across firm boundaries as individuals
seek out help from their external professional communities. The firm then develops
the knowledge as to how to coordinate these different professional individuals.
The third implication from our findings is that the building of the capability to
manage these knowledge flows presents considerable challenges for a company’s
management. First as we found in our research, on-time performance was negatively
correlated to the use of external codified sources, yet positively correlated to the use of
internal codified sources. Too much external knowledge leads to missed deadlines and
overrun budgets, while the reuse of internal knowledge leads to on-time performance.
Based on our qualitative findings, the aspect of prestige including the “not-invented-
here” syndrome – the desire to develop one’s own solutions rather than reuse existing
solutions – and Zipf’s Law of Least Effort (1949) play a significant role in the choice
of knowledge sources.
While it was found that this matter of prestige was strong within Icon, it may
have its roots outside of Icon’s borders in the global community, primarily among the
programmers. One programmer explained that he started working on one project
because he really wanted to show Silicon Valley that other areas of the world could
produce “bleeding edge” products as well. While on the one hand, programmers were
inspired to make Icon the world’s best company, on the other hand, programmers were
pressured by their global community to produce the latest “cool” solution. In addition,
programmers were under a form of social pressure from their external community to
help fellow members solve their difficult problems, often attempting to “show off” in
front of the others. This was found to lead to conflicting goals or loyalty for the
programmers: best company vs. best profession (see figure 2). Creating a “cool”
solution or trying to impress a global community through solving another external
member’s difficult problem leads to longer hours worked, using unnecessary resources
as well as causing delays in product delivery to the customer. However, it is this
communication with communities that span organizational boundaries that leads to the
cross-fertilization of communities that then fosters creativity. And it is this creativity
that is essential for the continuous creation of a firm’s competitive advantage.
Figure 2 An Individual’s Conflicting Loyalties
Professional
Community of practice
in organization
lty
ya
Lo
Lo
ya
lty
Individual
Thus, the challenge for management is then to be able to align the use of both
the internal and external knowledge sources with the company’s competitive strategy.
If the company is pursuing more of a knowledge creation over a knowledge reuse
strategy (Hansen et al. 1999), then a greater use of external sources over internal
sources ensures creativity and the access to the latest solutions. However, if the
strategic focus is on knowledge reuse, then too much external use leads to an
inefficient use of resources. This is no new challenge, merely the exploration vs.
exploitation balance in a new setting. What perhaps has intensified this challenge is
that as internet technology develops so rapidly, management may have difficulty in
keeping abreast of developments, making it a challenge to know whether employees
are working on necessary value-adding activities. One manager summed up this
situation with reference to the programmers. “Programmers take us (management)
hostage. We never know whether they’re working on extra bells and whistles to
impress their buddies or whether it’s really a value-adding activity for the customer.”
342 ARTICLE THREE
A second challenge for management is that when employees are active in their
external communities, they are often involved in disclosing proprietary company
knowledge to other external community members. As stated above, an unwritten code
of conduct with fellow community members exists that includes reciprocity. In order
for members to gain knowledge, they must provide knowledge to others. Those who
do not give are cut off from the knowledge flows. In many cases, individuals pass
right over firm policies prohibiting such action as indicated in this quotation by one
programmer, “We pass over the nondisclosure agreements of different companies all
the time and trade company secrets.” Thus, management must be aware that
knowledge is leaking through the boundaries of the firm to the external world through
participation in these electronic communities. Whether or not this leakage dilutes the
firm’s competitive advantage is an area for further research although previous research
has indicated that the greater the trading of information across company boundaries,
the higher the firm’s relative performance (Schrader 1991). Again, what we are seeing
at Icon is no new phenomenon (see Mansfield 1985, Schrader 1991, von Hippel 1987,
and von Hippel & Schrader 1996); however, the ease with which this knowledge
leakage can occur has been greatly facilitated with the spread of the internet.
Finally, a third challenge is that when management hires a person, management is also
“hiring” the employee’s network as well. Thus, management must consider the
potential employee’s external network and how active this person is in his or her
network. If the person is very active in his or her external network, then the
individual’s time may be spent on external activities. As shown above, this can lead to
both positive and negative results for the company.
Thus, this research has provided us with several areas that require
management’s attention in a rapidly growing company in order to facilitate the
creation of competitive advantage based on knowledge. First, a clear knowledge
management strategy (e.g., knowledge creation vs. knowledge reuse) should be
communicated. Secondly, a high number of socialization activities is an important
factor in ensuring the creation of communities that are aligned with the company’s
corporate and knowledge management strategy. As individuals join the firm at a rapid
pace, this socialization facilitates the creation of and membership in communities of
practice. In addition, socialization promoting alignment between the culture of the
communities and that of the firm enables individuals to conduct themselves in the
company’s interests when trading information across company boundaries. Finally,
management should focus on building a critical mass within the knowledge
management systems in the company’s intranet, whether it be codified documents or
discussion forums. Communication of the importance of the company’s knowledge
management systems and the use of the intranet will then further improve usage.
In terms of the limitations of this study, we acknowledge that there is a need to
look at more than one firm and preferably with a larger sample of respondents before
coming up with any definitive conclusions. The questionnaire suffers from common-
method bias, so ideally we would also complement some of our measures with
secondary data on e.g. meetings attended, emails sent, hours on the web. But such a
data-collection process would be extremely time-consuming and difficult to arrange.
Finally, it is important to acknowledge that our choice of communities of
practice as our theoretical lens has its drawbacks. As noted several times, it is almost
impossible to define communities of practice in an operational way, so one ends up
falling back on measuring individual level patterns of interaction. And having moved
in that direction, there are a number of other theoretical angles that could and perhaps
should be incorporated, such as the vast literature on groups, environmental scanning,
and organizational cognition. These are issues that will be considered in future
research.
ACKNOWLEDGEMENTS
I would like to acknowledge the extremely valuable help I received from Dr. Julian
Birkinshaw in performing this research. In addition I would like to thank the
management at Icon Medialab who facilitated the data collection and encouraged me
during this fascinating research. Finally, I would like to thank Andy Schenkel and
Joachim Timlon and everyone at IIB for their encouragement and comments as well as
Ilkka Tuomi for his very insightful comments on a previous draft. Funding was
provided by the CAMINO project at the Institute of International Business at the
Stockholm School of Economics, whose generosity is greatly appreciated.
REFERENCES
Alavi, M. & Leidner, D.E. 1999. Knowledge management systems: Issues, challenges,
and benefits. Communication of the Association for Information Systems, 1:1-28.
Aldrich, H.A., & Fiol, C.M. 1994. Fools rush in? The institutional context of industry
creation. Academy of Management Review, 19,4: 645-670.
Bartlett, C. & Ghoshal, S. 1989. Managing Across Borders: The Transnational
Solution. Cambridge: Harvard Business School Press.
Boland, R.J. Jr. & Tenkasi, R.V. 1995. Perspective making and perspective taking in
communities of knowing. Organization Science, 6,4: 350-372.
Brown, J.S. & Duguid, P. 1991. Organizational learning and communities of practice.
Review, 40, 3: 90-111.
usefulness of electronic weak ties for technical advice. Organization Science, 7,
119-135.
Davenport, T.H. & Prusak, L. 1998. Working Knowledge. Boston: Harvard Business
School Press.
Davidsson, P., Lindmark, L., & Olofsson, C. 1996.. Näringslivsdynamik under 90-
talet. Stockholm: Nutek.
344 ARTICLE THREE
DeGeus, A.P. 1988. Planning as learning. Harvard Business Review, March-April: 70-
74.
Delmar, F., & Davidsson, P. 1998a. High-growth SMEs and Employment: Assessment
of best practice policies (DSTI/IND/PME(98)14). Paris: OECD.
Delmar, F., & Davidsson, P. 1998b. A taxonomy of high-growth firms. In P. D.
Reynolds, W. D. Bygrave, N. M. Carter, S. Manigart, C. M. Mason, G. Dale Meyer,
& K. G. Shaver (eds.), Frontiers of Entrepreneurship Research. Babson Park, MA:
Arthur M. Blank Center for Entrepreneurship, Babson College: 399-413.
Doz, Y., Asakawa, K., Santos, J.E.P, & Williamson, P.J. 1997. The metanational
corporation. Working paper, Insead, Fontainebleau.
Drucker, P. 1990. The emerging theory of manufacturing. Harvard Business Review:
94-102.
Edmondson, A. 1999. Psychological safety and learning behavior in work teams.
Administrative Science Quarterly. 44: 350-383.
Faraj, S. & M. McClure-Wasko 1999. The web of knowledge: An investigation of
self-organizing communities of practice on the net. Working paper, University of
Maryland.
Forbes. 1998. November.
Fulk, J. & DeSanctis, G. 1995. Electronic communication and changing organizational
forms. Organization Science, 6, 337-350.
Hagel, J. & Armstrong, A.G. 1997. Net Gain: Expanding Markets through Virtual
Communities. Boston: Harvard Business School Press.
Hansen, M., Nohria, N., & Tierney, T. 1999. What’s your strategy for managing
knowledge? Harvard Business Review, Jan/Feb.
Heneman, H.G. 1974. Comparisons of self and superior ratings of managerial
performance. Journal of Applied Psychology, 59, 5: 638-642,
Hedlund, G. 1994. A model of knowledge management and the N-form corporation.
Strategic Management Journal, 15
Hedlund, G. & Nonaka, I. 1993. Models of knowledge management in the West and
Japan. In P. Lorange, B. Chakravarthy, J. Roos and A. Van de Ven, (eds.),
Implementing Strategic Process. Blackwell Business: 117-145.
Hinds, P. & Kiesler, S. 1995. Communication across boundaries: Work, structure, and
use of communication technologies in a large organization. Organization Science,
6: 373-393.
Kogut, B. & Zander, U. 1992. Knowledge of the firm, combinative capabilities and
the replication of technology. Organization Science, 3,3: 383-397.
Kogut, B. & Zander, U. 1995. Knowledge of the firm and evolutionary theory of the
multinational corporation. Journal of International Business Studies, 26: 625-644.
Lave, J. & Wenger, E. 1991. Situated learning: legitimate peripheral participation,
Mansfield, E. 1985. How rapidly does new industrial technology leak out? Journal of
Industrial Economics, 34,2: 217-223.
March, J.G. 1991. Exploration and exploitation in organizational learning. Organization
Science, 21:71-87.
Nonaka, I. 1991. The knowledge-creating company. Harvard Business Review,
November-December: 96-104.

Nonaka, I. & Takeuchi, H. 1995. The Knowledge-creating Company. New York:
Oxford University Press.
Orr, J. 1990. Talking about Machines: An Ethnography of a Modern Job. Cornell
University.
Purser, R.E., Pasmore, W.A., & Tenkasi, R.V. 1992. The influence of deliberations on
learning in new product development teams. Journal of Engineering and
Schein, E.H. & W. Bennis. Personal and Organizational Change via Group Methods.
New York: Wiley.
Schön, D.A. 1983. The Reflective Practitioner: How Professionals Think in Action.
New York: Basic Books, Inc.
Sjöberg, L., & Lind, F. 1994. Work motivation in financial crisis: a study of prognosis
factors. Stockholm School of Economics, Stockholm, Sweden.
Snyder, W. 1996. Organizational Learning and Performance: An Exploration of
Linkages Between Organizational Learning, Knowledge, and Performance.
University of Southern California. Unpublished doctoral dissertation.
Snyder, W. 1997. Communities of practice: combining organizational learning and
strategy insights to create a bridge to the 21rst century. Boston: Academy of
Management Conference.
Spender, J.-C. 1989. Industry Recipes: The Nature and Sources of Managerial
Judgment. Oxford: Basil Blackwell.
Spender, J.-C. 1996. Making knowledge the basis of a dynamic theory of the firm.
Strategic Management Journal. 17(winter special issue): 45-62.
Spender, J.-C & Grant, R.M. 1996. Knowledge and the firm: An overview. Strategic
Sproull, L.S. & Faraj, S. 1995. Atheism, sex and databases: The net as a social
technology. In Keller, B.K.J. (ed.) Public Access to the Internet. Cambridge, MA:
MIT Press.
Stebbins, M.W. & Shani, A.B.R. 1995. Organization design and the knowledge
worker. Leadership & Organization Development Journal, 16: 23-30.
Teigland, R. 1999. Netovation: Linking cross-boundary knowledge flows to firm
innovative performance in the internet industry. Working Paper 1999. Institute of
International Business, Stockholm School of Economics.
Wenger, E. 1998. Communities of Practice: Learning, Meaning, and Identity.
Wenger, Sep 17, 1997. Community of Practice Forum. Boston. 1997.
Wexley, K.N., R.A. Alexander, J.P. Greenawalt, and M.A. Couch. 1980. Attitudinal
congruence and similarity as related to interpersonal evaluation in manager-
subordinate dyads. Academy of Management Journal, 23,2: 320-330.
Venkatraman, N. & Henderson, J.C. 1998. Real strategies for virtual organizing. Sloan
Management Review, 40: 33-45.
Zander, U. 1991. Exploiting a Technological Edge – Voluntary and Involuntary
Dissemination of Technology. Published doctoral dissertation. Institute of
International Business, Stockholm, Sweden.
346 ARTICLE THREE
Zipf, G.K. 1949. Human Behavior and the Principle of Least Effort. Cambridge, MA:
Addison-Wesley.
APPENDIX
Table 7 Descriptive Statistics and Correlations for All Variables a

Variable Unit Mean s.d. 1 2 3 4 5 6 7 8 9 10
Self-evaluated performance
1. Creativity 1-7 scale 5.06 1.04
2. On-time performance 1-7 scale 4.96 1.22 .08
Control
3. Education 1-5 scale 2.98 1.03 .10 -.15*
4. Related work experience years 4.76 4.42 .29*** .20* -.04
5. Employed at Icon days 385.52 349.92 .18* -.08 .03 .11
6. Office openness 1-7 scale 5.13 0.63 .27*** .18** -.06 .05 .16*
External
7. Codified 1-7 scale 3.63 1.21 .17* .00 -.00 .13 .02 .05
8. External community interaction 1-4 scale 2.79 0.84 .21** .04 .07 .17 .14 .14 .13
Internal
9. Codified 1-7 scale 3.68 1.34 .06 .23** .04 .18 -.14 .13 .15* .17*
10. Internal community interaction 1-4 scale 3.12 0.62 .15* .15* .11 .02 .09 .29*** .10 .20** .08
11. Social contact 1,no; 2, 1.36 0.41 .27** -.06 .05 -.08 .10 .24** .12 .14 .09 .27**
yes
a Range of N is 154-203.
* p < .05
** p < .01
*** p < .001
Article Four
Extending Richness with Reach: Participation and

Knowledge Exchange in Electronic Networks of
Practice
In general, the terminology in this article corresponds to the terminology

in this thesis. The only concept worth noting is “General Performance”,
which is neither the same as Creative Performance or Efficient
Performance in the other articles. General performance does, however,
relate to a high level of expertise and is primarily related to creativity. In
addition, we use two abbreviations:
1) CoP for Community of practice

2) ENoP for Electronic network of practice
350 ARTICLE FOUR

Innovation Through Communities of Practice, London: Idea Group Inc., 2004
(forthcoming).
Earlier version published in W.J. Orlikowski, S. Ang, P. Weill, H.C. Krcmar, &
J.I. DeGross (eds.), The Proceedings of the 21st International Conference on
Information Systems (ICIS), Brisbane, Australia, 2000: 313-328.
Extending Richness with Reach: Participation and

Knowledge Exchange in Electronic Networks of
Practice
Robin Teigland
Sweden
Molly Wasko
ABSTRACT
In an effort to replicate communities of practice online, organizations are
investing in information technologies that create intra-organizational electronic
networks, or “electronic networks of practice”. These networks are designed to
enable the creation of electronic “bridging ties” between geographically
dispersed organizational members to provide a communication space in which
individuals working on similar problems may quickly ask each other for help
on task-related problems. This article compares the dynamics of knowledge
exchange between electronic networks of practice and traditional communities
of practice. In addition, this article examines why people participate and help
others in an electronic network of practice as well as whether electronic
network of practice participation has an impact on knowledge outcomes and
individual performance. In order to investigate these issues, data were
collected from a successful electronic network at one of Europe’s largest
consulting companies. The article concludes with a discussion of the results
and implications for both managers and researchers interested in the dynamics
of electronic knowledge exchange.
Keywords: knowledge management, electronic community, community of

practice, internet, performance
EXTENDING RICHNESS WITH REACH 351
INTRODUCTION
Communities of practice (CoPs) are regarded as essential building blocks of the
knowledge economy and are being promoted within organizations as sources of
competitive advantage and facilitators of organizational learning. In
organizations, CoPs traditionally emerge through the mutual engagement in
work performed by individuals who are either physically co-located or who
frequently meet each other face to face (Orr, 1996; Wenger, 1998). However,
due to hypercompetitive conditions in the marketplace and the increasing
complexity and diversity of global organizations, knowledge workers engaged
in the same practice are increasingly becoming more distributed across an
organization’s geographical locations. Thus, in an effort to replicate traditional
CoPs electronically, management in numerous organizations has invested in
computer-mediated communication technologies to facilitate knowledge
sharing regardless of time and space constraints. We refer to these emergent
virtual communities as electronic networks of practice (ENoPs). We follow
Brown and Duguid (2000) in their use of the term “networks of practice”, yet
we add the term “electronic” to highlight that communication within this
network of practice occurs primarily through computer-based communication
technologies, such as bulletin boards, listservs, etc. In this article, we use the
terms electronic networks of practice, networks, and ENoPs interchangeably to
avoid repetition.
While traditional, face-to-face CoPs within organizations have received
increasing attention, we know much less about the dynamics underlying ENoPs
and the electronic knowledge exchange supported by these computer networks.
Initial research suggests that participation in these networks provides access to
useful sources of technical advice for organizational members (Constant,
Sproull, & Kiesler, 1996). However, there is ample evidence that simply
investing in information technologies does not directly enhance knowledge
sharing. In fact, researchers estimate that between 50-70% of knowledge
management projects fail to meet expectations and stated objectives and
attribute these failure rates to an over-reliance on information technology
(Ambrosio, 2000). Thus, a key question for researchers and managers alike is
how to turn an empty electronic space into a vital, active forum devoted to
knowledge exchange.
The goal of this article is to provide guidelines to both researchers and
managers interested in studying and supporting electronic networks of practice
within and across organizations. In order to do so, we begin by presenting the
key characteristics that define an ENoP and compare ENoPs to CoPs. We then
examine two questions related to individual participation in an ENoP: (1) why
do people participate and help others, and (2) does participation result in
positive knowledge outcomes? Finally, we present and discuss findings from a
352 ARTICLE FOUR
recent study that investigated the above two questions in a successful ENoP at a
global consulting organization.
ELECTRONIC NETWORKS OF PRACTICE

The concept of a community of practice has highlighted the importance of
emergent mutual engagement in practice, where mutual engagement typically
refers to physically co-located, face-to-face interactions. However, emergent
mutual engagement can also occur through text-based communication and
discussion fora, such as bulletin boards, listservs and Usenet Newsgroups.
Thus, electronic networks of practice are similar to communities of practice in
that they are a social space where individuals working on similar problems self-
organize to help each other and share perspectives about their practice.
However, unlike a CoP, in an ENoP mutual engagement occurs through
computer-mediated communication. This profoundly affects how knowledge is
exchanged in several ways.
First, as mentioned, similar to a CoP, knowledge is exchanged in an
ENoP through mutual engagement in practice. Thus, one defining
characteristic is that participants in an ENoP interact with one another to help
each other solve problems. By posting a message to the network, individuals
requiring help with a problem may quickly reach out to other participants that
then provide valuable knowledge and insight in response. The network also
provides a forum for participants to share stories of personal experiences and
discuss and debate issues relevant to their practice (Wasko & Faraj, 2000). The
posting and responding to messages is recorded like a conversation between
participants, representing active mutual engagement in problem solving. This
characteristic of mutual engagement distinguishes ENoPs from more static
forms of electronic knowledge exchange, such as document repositories and
other types of databases.
Second, knowledge in an ENoP is exchanged through asynchronous,
text-based computer-mediated communication. In face-to-face interaction,
participants perceive a variety of social and visual cues, and have access to
immediate feedback. However, in electronic communication these cues are
filtered out, making it a lean medium of exchange and impacting how
knowledge is actually exchanged between participants (Daft & Lengel, 1986).
In addition, the technology creates a weak structural link between like-minded
individuals who are physically dispersed, thus eliminating the need for people
to know one another personally in order to access knowledge. In an ENoP, the
technology supports any number of participants, eliminating constraints due to
size. Therefore, knowledge seekers are not limited to asking only experts
whom they personally know or are able to identify, increasing the likelihood of
connecting with someone willing and able to help. Finally, mutual
engagement in an ENoP is typically archived and available to all participants in
the network. This creates an on-line repository of questions and answers that
can be referred to at a later time by any interested individual, regardless of his
or her participation in the original engagement or tenure in the electronic
network of practice. This contrasts with knowledge exchange in a CoP where
access to advice is limited to whom you know, and knowledge is exchanged
between seeker and provider without necessarily being made available to other
members of the community.
Another defining characteristic of an ENoP is that participation is open
to anyone with a desire to interact. The electronic links created by internet and
intranet technologies that enable individuals to communicate are practically
ubiquitous, thus membership is available to anyone with a connection. In
addition, because membership is open, membership is fluid, making it difficult
to create and enforce boundaries. This sharply contrasts with the tightly knit
relationships between specific members that typify CoP structures. Also, this
characteristic separates an ENoP from a virtual group or team, where members
are designated and assigned.
Fourth, participation in an ENoP is voluntary. Individuals choose
whether or not they want to participate as well as how often they participate -
ranging from simply lurking to becoming an active participant. In addition,
individuals have choices about how they participate, deciding whether or not to
post questions, replies, or both. Finally, individuals voluntarily determine what
they want to contribute, choosing what knowledge they are willing to disclose
as well as the length of the messages they contribute, influencing the quality
and helpfulness of the knowledge exchanged. This criterion of voluntary
participation distinguishes an ENoP from other forms of virtual work, such as
virtual teams, where participants are expected to coordinate efforts to deliver a
specific outcome.
Finally, participants in an ENoP are typically strangers. Because access
to the technology is practically ubiquitous, there are basically no limits to size,
and these networks are open to anyone, knowledge exchange occurs between
people regardless of personal acquaintance, familiarity and location. Also,
because participation is voluntary, a knowledge seeker has no control over who
responds to their questions or who uses their responses. This sharply contrasts
with a CoP where people typically know one another and interact over time,
creating expectations of obligation and reciprocity that are enforceable through
social sanctions.
We now turn to the two questions we raised above regarding
participation in an ENoP: (1) why do people participate and help others in an
ENoP, and (2) does participation in an ENoP result in positive knowledge
outcomes.
354 ARTICLE FOUR
Investigating Participation in an Electronic Network of Practice

Mutual engagement in an ENoP is open, voluntary and results in the creation of
a knowledge repository of archived messages that is available to all individuals
regardless of their original participation. Thus, one helpful theoretical lens
with which to investigate ENoPs is the theory of collective action and public
goods. A public good, for example a public park, is a resource that is created
only if a group of individuals contribute towards its production. However, a
public good cannot be withheld from any member of the collective, even if he
or she does not participate in the production or maintenance of the good
(Samuelson, 1954; Olson, 1965). With public goods, the optimal individual
decision is to enjoy the public good without contributing anything to its
creation or maintenance and to simply free-ride on the efforts of others.
However, if everyone were to act rationally and decide not to contribute, then
the good would never be created and everyone would be worse off.
ENoPs are a type of collective in which the knowledge exchanged and
created is the collective’s public good. As discussed above, mutual
engagement in an ENoP is open and voluntary. Participation typically results
in the creation of a knowledge repository of archived messages that is available
to all individuals regardless of their original participation. This begs the
question then – why would anyone invest their valuable time and effort helping
strangers in an ENoP if it is in their best interest not to do so?
Does Participation in an Electronic Network of Practice Affect

Knowledge Outcomes?
Another important issue to investigate is whether ENoPs exhibit the same
degree of continuous incremental innovation as CoPs. As mentioned above,
CoPs are generally characterized by rich, face-to-face exchange through
person-to-person interactions. Mutual engagement between individuals in a
CoP creates boundaries around the shared practice within which the
community’s knowledge is embedded, and tacit knowledge is shared relatively
easily between individuals within the community, often without ever being
made explicit. These tightly knit social structures facilitate the creation of a
shared identity through the development of a common language, social capital
(such as norms, trust, and obligation), boundaries, and social controls, resulting
in strong social ties between individuals. These characteristics have been
argued as essential for the continuous incremental improvements in the
community’s practice and the reason why CoPs are centers for learning and
innovation within organizations (Wenger 1998, Brown & Duguid 1991, 1998).
In contrast, interactions in an ENoP are limited to text-based,
asynchronous, computer-mediated communication. As a result, the ability of
members to develop a shared identity and common language through narration,
collaboration, and social construction is hampered. However, ENoPs have a
greater reach than CoPs, supporting the creation of weak electronic “bridging
ties” between an unlimited number of like-minded others. Due to the extensive

reach of these networks, individuals benefit from ENoPs since they gain access
to new information, expertise, and ideas that are often not available locally. As
such, the weak tie relationships created in an ENoP potentially increase an
individual’s access to greater resources and advice than are available in the
local community. Thus, one question to ask is whether this extended reach
results in positive knowledge outcomes. In other words, are weak electronic
links, like their strong tie counterparts, also useful for supporting knowledge
sharing and innovation?
EMPIRICAL STUDY
This study was undertaken in the Nordic operations (Denmark, Finland,
Norway, and Sweden) of Cap Gemini and was performed prior to the merger of
Cap Gemini and Ernst & Young Consulting. As a result, the company
description considers only the Cap Gemini organization. Cap Gemini is
Europe’s largest IT services and management consulting company, and within
the Nordic region, Cap Gemini has numerous networks designed to enhance the
company’s knowledge management activities. We chose participants in one
electronic network, which was referred to as the NCN MS Community. This
electronic network had 345 members spread across the Nordic countries, and
the members of this network all worked with applying Microsoft products in
their responsibilities with Cap Gemini. In order to communicate with each
other, a listserv was created that was nicknamed the L2A2L mailing list. The
nickname was based on the slogan “Learn to Ask to Learn” that was developed
to encourage knowledge sharing within this network. Network members
primarily used the L2A2L mailing list when they had a question regarding how
to perform their tasks at work. Thus, when one person needed help, he or she
posted a question to the whole network through the listserv. At the time of the
data collection, there were between five and ten requests for help per day on the
L2A2L mailing list.
Data were collected through the use of a web-based questionnaire that
was sent as an email attachment to each of the NCN MS Community members
during January 2000. Of the initial 345 individuals with valid email addresses,
we received a total of 83 usable survey responses for a response rate of 24%.
These 83 indicated that they had developed on average personal ties with 2.8
other members through participation in the listserv. The average age of the
respondents was 35.6 years with an average of 4.0 years employed at Cap
Gemini and 7.7 years of experience in their competence. The sample was 8%
women. Consultation with Cap Gemini’s management indicated that the
demographic characteristics of the group of respondents were representative of
those of the entire NCN MS Community. Specific variables were assessed
356 ARTICLE FOUR
through survey responses, and participants also provided insights by

responding to open ended questions about their participation.
Survey Results
In order to assess why people participate, and whether or not participation
results in positive knowledge outcomes, we posed four open-ended questions to
the participants in network: (1) why do you participate in the NCN MS
Electronic Community, (2) why do you help others with their problems, (3) has
your participation improved your work performance, and (4) how can the NCN
MS Community be improved? The following provides a summary and a
discussion of the results.
1) Why Do You Participate in the NCN MS Electronic Community?

In response to this question, respondents indicated that the ENoP was an
excellent means of improving their own level of technical competence.
Individuals responded that they learned through their participation by receiving
help and information related to their work tasks through participation.
Additionally, they felt that participation enabled them to keep up-to-date with
technical developments as well as to know who was actively working in
different areas. One respondent summarized the above in the statement,
“There is so much to know in this field and new applications/methods, etc. are
introduced all the time. I never know when I need this new knowledge in my
daily work or for a new project. The only thing I know is that I must always
learn new things!”
2) Why Do You Help Others with Their Problems?

There seem to be a variety of reasons why people take time out from solving
their own problems to help others in the ENoP. From the answers provided, it
appears that a norm of reciprocity developed between the members of the
network. In other words, in order to receive help from the network, individuals
felt obligated to help others in return. One individual explained, “It’s the way it
is! I help them; they help me in return.” In this manner, individuals felt that
they ensured that their individual competence level remained competitive, as
one respondent wrote, “Why shouldn’t I participate - knowledge devaluates
over time. Who likes to sit alone back with yesterday’s knowledge?” In
addition, respondents felt that helping others was a part of their job at Cap
Gemini. Through their participation, they were able to improve the level of
technical competence of the network as a whole and ultimately Cap Gemini’s
ability to be competitive in the market. One respondent summarized this in the
following quotation, “At the end of the day, we are a knowledge company. My
company moves faster the more knowledge can flow freely inside. This
knowledge flow will result in happier customers and more business.”
3) Has Your Participation in the NCN MS Electronic Community Improved

Your Work Performance?
The third question focused on investigating whether participation in the listserv
had helped improve work performance. Of the 83 respondents, 65% replied
that participation in the network had helped them. Two categories of answers
were provided. First, participation greatly improved the speed with which
participants were able to solve their problems. One individual replied, “Yes, I
get answers to more complicated questions much faster than trying to find the
answer by myself.” Second, individuals were able to learn and receive new
insights from the network as one commented, “Yes. I learn things from every
topic. Even when I am not working within the actual topic.”
4) How Can the NCN MS Electronic Community Be Improved?

Finally, we asked about how the listserv could be improved. Suggestions
included creating a database of all the postings such that individuals could
easily find previously discussed topics as well as a means to educate people on
how to use the list. There also seemed to be a discrepancy in terms of the most
effective communication channel for this network. Several individuals
indicated that they would prefer a discussion forum to a mailing list since they
felt that the overall level of email received daily was too high. We discussed
this with the network head. However, he felt that one of the reasons that this
network did have such a high participation rate was due to the choice of a
mailing list over a discussion forum. He explained that a mailing list led to a
higher level of activity since participants automatically received postings in
their inbox while the use of a forum required that the participant actively enter
the discussion forum.
Summary of Survey Items

In addition to the open-ended questions, we also included specific survey items
to provide additional quantitative support examining the relationships between
participation in the network and knowledge outcomes. Summary statistics and
correlations are presented in table 1. The exact wording of specific items is
listed in Appendix 1.
Results indicate that higher levels of participation and tenure in the
ENoP are associated with both acquiring knowledge from participation in the
network and contributing knowledge to others. In addition, both knowledge
acquisition from, and knowledge contribution to the network are positively
related to individual performance. However, tenure in the ENoP is not
associated with higher rates of participation or with individual performance.
Finally, the results suggest that individuals who rely on their co-located
colleagues for help or advice with their work tasks report no associations with
participation, knowledge acquisition, or knowledge contribution. In fact, the
358 ARTICLE FOUR
survey results indicate that reliance on co-located colleagues is associated with

lower levels of self-reported individual performance.
Table 1 Quantitative Results from Survey
Scale Std. Cronbach’s

Range Mean Dev. Alpha 1 2 3 4 5
1. ENOP Participation
1-7 2.3 0.82 n/a
Level
2. ENOP Tenure 1-50 10.81 11.64 n/a .06
3. Knowledge Acquisition 1-7 3.62 1.75 0.95 .52** .29*
4. Knowledge Contribution 1-7 2.34 1.57 0.85 .59** .23* .52**
5. Co-located Coworkers 0-3.71 2.99 1.05 n/a .04 .06 .02 .08
6. Individual Performance 1-7 4.30 1.29 0.82 .41** .13 .24* .57** -.24*
* Significant at the p < .05 level, two-tailed
** Significant at the p < .01 level, two-tailed

This research indicates that people who participate and help others in this
ENoP are not acting irrationally. Rather, they choose to participate in order to
gain exposure to critical new ideas and to access help and advice not available
locally. In addition, another key dynamic underlying knowledge exchange in
this network is a strong norm of reciprocity. Individual participation is
sustained by a strong sense of paying back to the network by helping others in
return. Another key motivation underlying why people participate is related to
identification with the organization, or a strong sense of organizational
citizenship. Interestingly, the results suggest that the level of participation in
the network is more important for supporting positive knowledge outcomes
than the length of time an individual has participated. This implies that
newcomers to the network can reap the same benefits of participation as long
standing participants.
Results also indicate that characteristics of the communication
technology supporting the network are important. The two major technologies
supporting ENoPs are listservs and bulletin boards. One advantage to the
listserv technology is that it is delivered to participants via e-mail, which
people frequently check or are notified automatically when a new message is
received. Thus messages posted to the network are “pushed” to the participants
and made visible along with e-mail. However, participants in this ENoP
indicate that one disadvantage of this technology is that the messages are not
stored in a single repository that can be accessed as a FAQ for newcomers, or
searched for historical information. Bulletin boards are automatically arranged
in discussion threads, making it easier to archive and search prior interactions.
However, participants must voluntarily take the time to actually visit and
participate in the network.
Returning to the question as to whether a network of practice is able to

support continuous incremental innovation and positive knowledge outcomes,
both the quantitative and qualitative results suggest that individual members
did improve their individual performance through their network participation.
This finding indicates that computer-mediated communication may be
sufficient to support the complex interactions necessary for the combination
and exchange of knowledge between individuals, thus facilitating their ability
to learn. In addition, it appears that individuals value accessing new insights
and ideas through weak electronic links that transcend their strong tie networks.
There is also evidence to suggest that individuals who participate in an ENoP
outperform their colleagues who primarily rely on their co-located colleagues
for knowledge and advice.
This finding suggests that knowledge in a tightly knit CoP may be
largely redundant, providing little additional information over what an
individual may already know, thus impeding the ability to develop new and
creative ideas (Granovetter, 1973, 1983). In addition, while CoPs and reliance
on face-to-face interactions with coworkers supports knowledge exchange,
learning and innovation, it has also been suggested that tightly knit CoPs may
lead to the “not invented here” syndrome or the resistance to new ideas not
locally developed. Thus, managers concerned with improving knowledge
exchange should note that the highly efficient structures that support
knowledge integration in a CoP may evolve into core rigidities and competency
traps – inappropriate knowledge sets that preserve the status quo and limit new
insights (Levitt & March, 1988; Leonard-Barton, 1992). Our findings suggest
that one way to alleviate this concern is to use ENoPs to create electronic
bridging links between strong tie communities to enhance the flow of new
ideas and innovations.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support provided by Cap Gemini
Nordic and in particular that of Christian Forsberg , Christian Storck, and Carl
Anlér as well as Victor Sylvan and Hanna Janson of the Stockholm School of
Economics.
REFERENCES
Ambrosio, J. 2000. Knowledge management mistakes. Computerworld, 34,27:
44.
Brown, J.S. & Duguid, P. 1998. Organizing knowledge. California
360 ARTICLE FOUR
Brown, J. S., & Duguid, P. 2000. The Social Life of Information. Boston, MA:
usefulness of electronic weak ties for technical advice. Organization
Science, 7,2: 119-135.
Daft, R. L. & Lengel, R. H. 1986. Organizational information requirements,
media richness and structural design. Management Science, 32,5: 355-366.
Granovetter, M. 1983. The strength of weak ties: A network theory revisited.
Sociology, 91: 481-510.
managing new product development. Strategic Management Journal, 13
(Summer Special Issue): 111-125.
Levitt, B., & March, J. G. 1988. Organizational learning. Annual Review of
Sociology, 14: 319-340.
Olson, M. 1965. The Logic of Collective Action. Cambridge, MA: Harvard
University Press.
Orr, J. 1996. Talking About Machines: An Ethnography of a Modern Job.
Ithaca, NY: ILR Press.
Samuelson, P. A. 1954. The pure theory of public expenditure. Review of
Economics and Statistics, 36,4: 387-389.
Wasko, M., & Faraj, S. 2000. It is what one does: Why people participate and
Information Systems, 9, 2-3: 155-173.
Wenger, E. 1998. Communities of Practice. Cambridge, UK: Cambridge
University Press.
APPENDIX 1
1. Electronic Network of Practice Participation Level

How often do you participate in the NCN MS electronic community?
Never, I mostly lurk (reading without posting), 0-5 times per week, 5-10 times
per week, 10-20 times per week, More than 20 times per week
2. Electronic Network of Practice Tenure

How long have you been a member of the NCN MS electronic community?
___ months
3. Knowledge Acquisition
From your interaction in the NCN MS electronic community have you:
Acquired knowledge that caused you to develop 1, to a very small extent, 7,
new insights to a very great extent
Acquired knowledge that enabled you to perform 1, strongly disagree, 7,
new tasks strongly agree
4. Knowledge Contribution
From your interaction in the NCN MS electronic community have you:
Contributed new knowledge to the NCN MS 1, to a very small extent, 7,
electronic community to a very great extent
Contributed knowledge to other NCN MS 1, strongly disagree, 7,
electronic community members that resulted in strongly agree
their development of new insights
5. Reliance on Co-located Coworkers

How often do you use the below information sources in your everyday work?
1, to a very small extent, 7,
Coworkers in my location to a very great extent
362 ARTICLE FOUR
6. Individual Performance
Please rate the extent of your agreement with each statement using the scale
below:
Able to develop creative solutions relative to your 1, strongly disagree, 7,
colleagues at Cap Gemini strongly agree
I have a high level of expertise in the technology 1, strongly disagree, 7,
with which I work strongly agree
My colleagues at Cap Gemini consider me to be a 1, strongly disagree, 7,
guru strongly agree
Article Five
Integrating Knowledge Through Information-

Trading: Examining the Relationship Between
Boundary Spanning Communication and Individual
Performance
There are several terms worth explaining in this article. First, boundary
spanning communication is used to connote participation in various
networks of practice. Below is a further key to how the terminology in
this thesis corresponds to the terminology in this article.
Thesis Terminology Article Terminology

Knowledge trading, knowledge Information trading
sharing
Participation in networks of Boundary spanning
practice communication
Individuals Knowledge workers
Emergent Informal
Network of practice Informal social network
Electronic network of practice Electronic discussion network
Efficient performance General performance
364 ARTICLE FIVE

Management, 2003 (forthcoming)
Integrating Knowledge Through Information

Trading: Examining the Relationship Between
Boundary Spanning Communication and Individual
Performance
Robin Teigland
Institute of International Business, Stockholm School of Economics,
Stockholm, Sweden
Molly McLure Wasko

ABSTRACT
With the global penetration of internet technologies, individuals may now cross
organizational boundaries to communicate efficiently with others in various
networks of practice regardless of time and space. Thus, when looking for help
in solving work tasks, knowledge workers may just as easily contact
individuals in rival firms across the globe as a coworker sitting at the next desk.
As a result, management faces questions such as 1) how should firms manage
employees’ knowledge-sourcing activities when they span both intra-
organizational and extra-organizational boundaries, and 2) what is the
relationship between different knowledge-sourcing activities and individual
performance? Grounded in the knowledge-based view of the firm, we
investigate these questions using data from Europe’s largest IT services and
management consulting company. Our results provide evidence that
organizations should support boundary spanning and participation in both
internal and external networks of practice. Results suggest 1) a positive
relationship between boundary spanning communication and creativity and
general performance and 2) a negative relationship between a reliance on co-
located coworkers as knowledge sources and creativity.
Keywords: knowledge, network of practice, community of practice, know-how

trading, boundary spanning, performance, electronic community, structural
equation modeling
INTEGRATING KNOWLEDGE 365
INTRODUCTION
With the rapid penetration of internet communication technologies across the
globe, the possibility for individuals to seek out others for advice and know-
how has dramatically increased. Individuals may now cross organizational
boundaries to communicate efficiently with others regardless of time and space
(Hinds & Kiesler, 1995) and participate in networks of practice, or emergent
networks of relationships built on work-related interactions (Brown & Duguid,
2000). As a result, many organizations are in the process of implementing
intranet-based communication tools, such as electronic discussion networks, to
promote knowledge sharing across internal organizational boundaries (Fulk &
DeSanctis, 1995; Davenport & Prusak, 1998; Alavi & Leidner, 1999).
Investments in these technologies are driven by the assumption that knowledge
is the most valuable resource of the firm and that new knowledge is created
through the recombination and exchange of existing knowledge (Kogut &
Zander, 1992; Nonaka, 1994; Nahapiet & Ghoshal, 1998). This knowledge-
based view of the firm proposes that sustainable competitive advantage stems
from an organization’s ability to integrate tacit knowledge embedded in the
minds of individuals (Nonaka, 1994; Grant, 1996a). Thus, increasing the
amount of information sources and communication channels employees have
available should increase the likelihood of new knowledge creation, resulting in
an improved level of performance for both the individual and the firm.
However, in addition to facilitating intra-organizational knowledge
flows, the internet also enables individuals to quickly and effortlessly access a
wide variety of knowledgeable individuals outside company boundaries
through email or other informal means such as Usenet groups, private chat
rooms, electronic discussion networks, listservs, etc. Thus, individuals are able
to communicate and share advice with thousands of others across the globe
regardless of their demographic characteristics, organizational setting, or local
culture (Sproull & Faraj, 1995; Faraj & Wasko, 1998). Communication across
intra-organizational and extra-organizational boundaries has been researched
extensively - a major stream began in the 1960s with the investigation into the
communication patterns of scientists and engineers in R&D laboratories (Allen,
1977). However, what has changed within the past ten years is the ease and
speed with which employees at all organizational levels can participate in these
knowledge flows.
Sharing knowledge across internal and external organizational
boundaries poses novel challenges to organizations attempting to manage their
knowledge resources (Pickering & King, 1995). Through interaction with
external sources, individuals gain access to information and expertise not
available locally and can interact informally, free from the constraints of
hierarchy and local rules. However, interaction with these external sources
usually involves a high degree of information trading and reciprocity. In order
for individuals to receive help from external sources, they must be willing to
366 ARTICLE FIVE
give advice and know-how as well, some of which company management may
consider proprietary (Von Hippel, 1987). In addition, much of the prior
research on boundary spanning communication looked at the relationship
between these knowledge flows and team performance. Research that
examines the relationship between knowledge flows and individual
performance is scant, as is research on the resulting implications for
organizations concerned with managing knowledge assets.
Thus, the goal of this research is to examine whether the performance of
an individual knowledge worker varies as a result of boundary spanning
communication activity and informal information trading across intra and
extra-organizational boundaries within networks of practice. Specifically, we
examine whether individual performance is related to informal information
trading and accessing knowledge from: 1) co-located coworkers, 2) coworkers
within the same organization but located across intra-organizational boundaries
(non-co-located), 3) intra-organizational electronic discussion networks, 4)
informal contacts in other organizations (i.e., contacts that are not the result of
a formal relationship with the firm such as a customer, alliance partner,
supplier, or other formal relationship), or 5) inter-organizational electronic
discussion networks.
Such inquiry makes three important contributions. First, this research
empirically examines the trade-offs between accessing and applying local
knowledge and accessing knowledge through boundary spanning
communications in various networks of practice. Second, this research clarifies
how the use of various sources of advice is related to individual performance in
complex knowledge environments. Finally, this research makes possible more
precise theoretical models of how internet-based communication technologies
can be designed and deployed to support knowledge exchange and the creation
of new knowledge to enhance individual and thus organizational performance.
Recent advances in strategic management thought suggest that organizational
resources and capabilities rather than served markets are the principal source of
sustainable competitive advantage, and that knowledge is the most important
strategic resource of the firm (Grant 1996a). This increased emphasis on
organizational capability and knowledge has led to the development of the
knowledge-based view of the firm (Grant, 1996a,b; Spender, 1996). Assuming
that knowledge is a critical input to production processes, then organizational
capability stems from the ability to integrate the specialized knowledge of
individuals (Nonaka, 1994; Grant, 1996a,b; Spender, 1996). Therefore, one of
the key issues underlying the knowledge-based view of the firm is to
understand how knowledge is integrated in firms to create organizational
capability (Hansen, 1996). However, since knowledge creation is based on
individual activity, the examination of knowledge integration requires
understanding the organizational processes through which firms access and
utilize the knowledge possessed by its members (Grant, 1996b).
According to Grant (1996a), competitive advantage results from how
effective firms are in integrating the specialized knowledge of their members,
and he proposes that this effectiveness depends upon the efficiency, the scope,
and the flexibility of knowledge integration. Efficiency refers to how
productive firms are in integrating individuals’ specialized knowledge. The
scope of knowledge integration refers to the different types of specialized
knowledge being integrated – the more complex the scope, the greater the
difficulty for competitors to replicate. The flexibility of integration reflects
extending existing capabilities through boundary spanning activities in order to
access and reconfigure additional knowledge through both internal and external
integration (Grant, 1996a). This research focuses on two aspects of Grant’s
theory of knowledge integration: efficiency and flexibility (Grant, 1996a).
As mentioned above, efficiency refers to how productive firms are in
integrating individuals’ specialized knowledge. One condition of integrative
efficiency is a common language of discourse to ensure efficient
communication between individual specialists. Efficient integration is also
dependent upon the frequency and variability of task performance, where
higher levels of frequency engender automated responses from each
organization member. Finally, organizing structures influence integrative
efficiency. Organizing activities that reduce the intensity and extent of
communication are needed to support efficient integration, such as modularity
and division of labor.
In addition to integrative efficiency, Grant also emphasizes the
importance of integrative flexibility. Hypercompetitive conditions in the
marketplace drive the eventual erosion of all positions of competitive
advantage. Thus, sustaining a competitive advantage requires flexibility and
the creation of new capabilities. Firms need to establish knowledge integration
techniques that extend existing capabilities by bringing in new knowledge and
reconfiguring existing knowledge. However, the need to access new
knowledge creates complex organizational issues with regard to firm structure,
firm boundaries, and choices between internal and external organizational
boundary spanning.
Grant’s theory of knowledge integration represents a paradox:
increasing the efficiency of knowledge integration may hinder flexibility and
the ability to create new innovations. For example, prior research suggests that
creating organizational structures that increase the efficiency of knowledge
integration through common language and frequent interactions may result in
knowledge hoarding, less creativity and the “not invented here syndrome”
(Granovetter, 1973; Szulanski, 1996). In addition, Grant’s theory focuses
primarily on the issue of coordination (structuring to enhance the effectiveness
of knowledge integration), without referring to issues of “cooperation”. The
theory leaves out a key component by assuming that people are willing to share
368 ARTICLE FIVE
knowledge openly and freely if provided with the structures/opportunities to

interact.
However, prior research on boundary spanning communication activities
within networks of practice suggests that individuals do not give away help and
advice to others in their informal social networks for free. Rather, individuals
trade knowledge with expectations of reciprocity. This informal trading
activity has been coined informal know-how trading and as originally
conceived, the concept did not distinguish between the types of knowledge
exchanged. In other words, know-how trading simply referred to the trading of
any type of information. Recent advances in knowledge management thought
suggest that know-how refers primarily to the more tacit procedural types of
knowledge, as distinguishable from know-what or the more explicit declarative
types of knowledge (Kogut & Zander, 1992). Thus, in order to stay consistent
with current thought, we refer to these informal reciprocal communication
exchanges as informal information trading rather than know-how trading. In
other words, the goal of this research is not to examine the type of knowledge
exchanged, rather to investigate the importance of reciprocity as a cooperation
mechanism. We propose that trading and expectations of reciprocity are the
key cooperation mechanisms underlying cross-boundary knowledge exchange.
Informal information trading between individuals has been shown to be
valuable and sustainable over time because the sharing of knowledge is an
important aspect of being a member of a professional community, even if the
employing organizations are direct competitors (Bouty, 2000). Therefore, key
issues for organizations interested in successfully managing their knowledge
resources involve understanding where knowledge workers turn for advice,
whether internally or across organizational boundaries, and exactly how they
access that advice. Previous research has indicated that reciprocal external
information trading involves “leakage”, or the flow of company proprietary
knowledge across firm boundaries (Mansfield, 1985; Von Hippel, 1987; Carter,
1989; Schrader, 1991). In previous research conducted by one of the authors
(Teigland, 2000), one programmer explained his communications in the
following manner:
“….but most importantly I have my network from the internet. I’ve

been in this for four years so really there is a core clique of people
who know each other and who trade secrets with each other. We
pass over the nondisclosure agreements of different companies all
the time and trade company secrets.”
This participation in information trading and the resulting potential for

informal proprietary knowledge flows across organization boundaries are of
particular strategic interest to management since such activity may impact a
firm’s competitive advantage. Yet, it is very difficult for firms to manage and
evaluate the benefits since it occurs “off the books” with employees generally
acting completely on their own with no managerial influence and no
documentation of the trade (Von Hippel & Schrader, 1996). In addition,
investigation into the relationship between this individual-level knowledge
exchange and performance has been extremely limited.
In summary, key strategic issues for firms and their managers are to better
understand how to balance efficient knowledge integration with demands for
flexibility, and how to manage the intra-organizational and extra-organizational
boundary spanning and informal information trading conducted by individual
knowledge workers. As a step in this direction, we develop a set of hypotheses
relating internal and external sources of knowledge, their influence on
information trading, and implications for individual performance.
HYPOTHESES AND MODEL DEVELOPMENT

In this section, we look at the drivers of internal and external information
trading and develop the relationships between information trading activity and
individual performance. In addition, we examine two types of performance: 1)
general performance or the ability to meet one’s job demands and 2) creativity
or the ability to develop creative solutions. We expect that the knowledge
sources used and the trading activities performed by individuals will impact
general performance and creativity in different ways, i.e., general performance
is reflective of application of current knowledge while creativity is concerned
with new knowledge creation and innovation.
Integrative Efficiency
Integrative efficiency requires common language, frequent interaction and
modular structure (Grant, 1996a). This suggests that one of the most efficient
sources of knowledge should be co-located coworkers who share the same
physical space since they are more likely to frequently interact with each other.
Ethnographic research on work practices finds that this frequent interaction
often occurs in communities of practice (Lave & Wenger, 1991; Orr, 1996;
Wenger, 1998). Researchers have found that these networks emerge over time
between individuals working on similar task-related issues located in face-to-
face settings. With knowledge-intensive tasks, often no one individual can
solve the problem on his or her own due to the inability to know everything.
Thus, when an individual becomes stuck in conducting a work-related task, he
or she often turns to knowledge sources that are the most easily accessed (such
as asking co-located coworkers), rather than searching for and using the best
knowledge source (such as codified sources or non-co-located coworkers)
(Gerstberger & Allen, 1968; O'Reilly, 1982). Through patterns of mutual
exchange and collaboration, individuals share knowledge to help each other
reduce the equivocality of problematic issues and build the community’s
memory (Orr, 1996; Wenger, 1998). Thus, individuals who rely on others in
their local setting to a high degree are likely to engage in a high degree of
mutual knowledge exchange.
370 ARTICLE FIVE
Through this mutual exchange and collaboration over time, individuals

become bound together by the context of the situation in an informal manner in
communities of practice, creating the social fabric of the organization (Brown
& Duguid, 1991). These emergent organizational structures provide the nexus
for the sharing and transfer of valuable individual and group tacit knowledge
(Kogut & Zander, 1992), resulting in higher performance of the community as
a whole (Brown & Duguid, 1991, 1998; Wenger, 1998). Individuals develop a
common language, explicit and tacit rules of behavior and coordination, and a
shared identity, (Wenger, 1998). Accessing knowledge from others who share
the same coding scheme and language is highly efficient (Tushman & Katz,
1980), thus economizing on the amount and intensity of communication needed
to achieve knowledge integration. In addition, in many work environments,
employees are confronted with information overload. Through asking someone
in the community for help, time does not have to be spent sorting though piles
of information for relevant documents (Wenger, 1998). This suggests that one
of the most efficient sources of knowledge should be co-located coworkers,
who are more likely to frequently interact with each other and develop into a
community of practice due to their sharing the same physical space. Thus, due
to the efficiency of integration, people who access knowledge from co-located
coworkers to a high degree should report higher levels of individual efficient
performance.
However, prior research argues that individuals within a social clique
tend to have strong ties, which have been defined as emotionally intense,
frequent, and involving multiple types of relationships, e.g., friends, advisors,
and coworkers (Granovetter, 1973). The result is that the knowledge held by
the members of a social clique tends to be redundant with that held by other
members, providing little additional information over what an individual may
already know (Granovetter, 1973, 1983). Thus, the knowledge available
through co-located coworkers is likely to be limited and superfluous, impeding
the ability to develop new and creative ideas. In addition, the highly efficient
structures that support knowledge integration and the exploitation of core
capabilities may evolve into core rigidities and competency traps –
inappropriate knowledge sets that preserve the status quo and limit new
insights, resulting in gaps between the knowledge of the firm and changing
market conditions (Levitt & March, 1988; Leonard-Barton, 1992). Therefore,
while accessing knowledge from co-located coworkers is likely to be highly
efficient and lead to better efficient performance, co-located coworkers are less
likely to offer the integrative flexibility needed to enhance creativity and
develop new capabilities. The above then leads to our first set of hypotheses:
HYPOTHESIS 1A: The greater the reliance on co-located coworkers to

access information, the lower the level of individual creativity.
HYPOTHESIS 1B: The greater the reliance on co-located coworkers to
access information, the higher the level of individual general
performance.
Internal Integrative Flexibility and Information Trading

Internal integrative flexibility involves the extent to which existing knowledge
within the firm can be reconfigured. While people tend to access knowledge
from those within the same physical proximity, advances in communication
technologies have made it easier for people who are working on similar task-
related problems across the organization to communicate. Thus, individuals
may relatively easily access knowledge from coworkers in their networks of
practice whom they personally know but who are working in other offices or on
site with clients. However, due to the less frequent patterns of interaction and
lower intensities of social pressure, non-co-located coworkers may be less
willing or committed to exchange knowledge without some type of return
(Blau, 1964). People within the same organization often prefer to hoard their
knowledge because they perceive that sharing knowledge results in reduced
status and lower levels of personal worth (Orlikowski, 1996). This is
especially evident in organizations where knowledge is the basis of a personal
competitive advantage over others (Thibaut & Kelley, 1959). Therefore,
individuals are more likely to expect reciprocity when engaging in knowledge
exchange across internal organizational boundaries, especially in situations
where relations are not characterized by frequent interactions and a high level
of trust (Nahapiet & Ghoshal, 1998). Thus, accessing advice and knowledge
from non-co-located coworkers is likely to result in internal information trading
with the expectation of reciprocity in return. This leads to our next hypothesis:
HYPOTHESIS 2: The greater the reliance on coworkers in other locations

to access information, the higher the level of internal information trading.
Informal social network structures connecting intra-firm acquaintances

have typically emerged through mutual engagement in work tasks, requiring a
personal, oftentimes physical, connection. However, recent advances in
computer-mediated communication technologies have facilitated the
development of intra-organizational, electronic social networks between
geographically dispersed organizational members, who are typically strangers.
Within these networks, an unlimited number of non-co-located, unacquainted
coworkers are able to quickly communicate through their shared organizational
and technical code to help each other solve problems and provide useful advice
(Constant, Sproull, & Kiesler, 1996).
However, theories of social capital suggest that the ability to develop the
commitment and trust that are necessary for knowledge exchange is difficult to
achieve in computer networks (Nohria & Eccles, 1992; Nahapiet & Ghoshal,
1998). Thus, norms of participation in electronic networks typically dictate
372 ARTICLE FIVE
that those who seek and receive help from the network must also pay back by
helping others (Constant et al., 1996; Kollock, 1999; Lakhani & von Hippel,
2000). Therefore, similar to the previous hypothesis, the sharing of
organizational knowledge through electronic networks is likely to increase the
amount of internal information trading within the firm. This leads to our third
hypothesis:
HYPOTHESIS 3: The greater the reliance on intra-organizational

electronic networks to access information, the higher the level of internal
information trading.
Through internal information trading, individuals exchange information,

knowledge and advice. These individuals share the same organizational
language and code of behavior and are faced with similar issues related to their
knowledge tasks, supporting integrative efficiency (Brown & Duguid, 1991;
Wenger, 1998). Internal information trading enables the flow of ideas and
innovations within the firm. For example, when seeking help internally, an
individual may find that a solution already exists elsewhere within the
organization. In this manner, individuals may avoid “reinventing the wheel” by
reconfiguring solutions previously developed within the firm to fit new
situations. These exchanges retain the context in which the knowledge is
embedded, and individuals located across intra-organizational boundaries
possess knowledge that may be more locally adapted, supporting efficient
integration and in turn, higher levels of general performance.
In addition, the combination and recombination of firm-specific
knowledge that is physically dispersed across the organization may facilitate
integrative flexibility. Individuals in other organizational units are more likely
than co-located coworkers to have important knowledge that is non-redundant,
generating access to sources of new ideas and innovations located across intra-
firm boundaries (Granovetter, 1973). Engaging in internal information trading,
people not only send and receive task-specific knowledge, they also help each
other by taking the time to work through each other’s problems. Exercising
intellect by helping others is likely to sharpen and even improve an individual’s
own technical skills. When an individual works through someone else’s
problems, he or she often discovers new methods and new applications for
existing knowledge (Wenger, 1998). Additionally, individuals that help others
are entitled to reciprocity, gaining access to advice, new ideas and innovations
when needed. Thus, we expect internal information trading to have a positive
impact on both individual creativity and general performance. This leads to the
following hypotheses:
HYPOTHESIS 4A: The higher the level of internal information trading, the
higher the level of individual creativity.
HYPOTHESIS 4B: The higher the level of internal information trading, the
higher the level of individual general performance.
External Integrative Flexibility and Information Trading

In addition to internal integrative flexibility, firms need to integrate new
knowledge found in the external environment to remain competitive (Cohen &
Levinthal, 1990). This knowledge may be accessed through either market or
relational contracts (Grant, 1996a). While relational contracts tend to refer to
formal inter-organizational arrangements, these contracts also comprise
informal communication exchanges between individuals. Just as intra-
organizational communication has been facilitated with the rapid spread of the
internet, so too has the ability to informally communicate with contacts outside
of the organization and to thus participate in inter-organizational networks of
practice (Brown & Duguid 2000). Thus, when knowledge workers seek help
with their work-related tasks, they may just as easily contact individuals
working in rival firms as individuals working in the same organization (Sproull
& Faraj, 1995; Faraj & Wasko, 1998). As a result, individuals may integrate
knowledge from within their organization with new ideas and innovations
accessed through communications with individuals outside their organization.
This informal relational activity should improve a firm’s external integrative
flexibility and sustainable competitive advantage (Grant, 1996a).
Informal information and knowledge sharing between firms has been
detected in several settings, e.g., semiconductor, specialty steel and mini-mill
industry, and R&D operations (von Hippel, 1987; Carter, 1989; Schrader,
1991). Reciprocity was found to be one of the guiding principles in these
informal exchanges, where individuals expected that their chances to receive
information in return would increase after they sent out information (Schrader,
1991). Certain professional disciplines encourage knowledge sharing and
information trading within the occupational community to keep abreast of new
ideas and innovations and to stay competitive with other professionals
(Pickering & King, 1995). As such, knowledge workers may also rely on their
informal external contacts, such as friends, family, ex-coworkers, people with
whom they attended school, etc., to access critical knowledge that resides
externally to the firm. However, due to strong norms of reciprocity within the
occupational community, relying on contacts in other organizations for advice
obligates the knowledge seeker to share knowledge in return (Van Maanen &
Barley, 1984). Thus, accessing knowledge from contacts in the occupational
community requires reciprocation through external information trading. This
leads to our next hypothesis:
HYPOTHESIS 5: The greater the reliance on external contacts to access

information, the higher the level of external information trading.
In addition to accessing information and know-how from external

contacts, communication tools such as bulletin boards, listservs, and chatrooms
374 ARTICLE FIVE
electronically connect knowledge workers sharing the same profession but who
are globally dispersed and typically strangers. In a recent article, Pickering &
King (1995) argued that the growth of inter-organizational internet-based
communication is likely to be especially rapid between individuals who are
interested in establishing ties with individuals outside of the firm based
primarily on similar professional interests.
In internet-based electronic networks, individuals are able to share
information and know-how through mechanisms that support posting and
responding to questions, sharing stories of personal experience, and discussing
and debating issues relevant to the professional community (Wasko & Faraj,
2000). Knowledge is continuously created and shared through open discussion
and collaboration, regardless of physical distance or organizational affiliation.
In one study of a Usenet inter-organizational technical discussion network, it
was found that 42% of all messages included programming code (Wasko &
Faraj, 1999). Thus, inter-organizational electronic networks advance the
knowledge of the professional community as a whole through electronic links.
Similar to accessing knowledge through external contacts whom an individual
knows, the norms of accessing knowledge from inter-organizational electronic
networks also requires that an individual “pay back” to the network by helping
others (Lakhani & von Hippel, 2000; Wasko & Faraj, 2000). Thus, we predict
the following:
HYPOTHESIS 6: The greater the reliance on external electronic

communities to access information, the higher the level of external
information trading.
There is limited empirical evidence that links external information trading

activities and performance. In one of the few studies in this area, Schrader
(1991) found suggestive evidence for a link between informal know-how
trading and firm performance, yet beyond this, there is little that investigates
the relationship to individual performance. However, research has found that
individuals who share the same professional interests may be able to
communicate relatively easily across organizational boundaries due to a
universal professional language, enabling individuals to access know-how and
information from outside the firm’s boundaries (Hauptman, 1986). This know-
how and information is more likely to be non-redundant than that found within
the organization. Thus, through external information trading, individuals may
combine and recombine knowledge from within their organization with new
ideas and innovations accessed from individuals outside the organization,
resulting in new and creative solutions (Cohen & Levinthal, 1990).
While individuals participating in external information trading share to a
certain extent the same professional knowledge and technical language, they do
not share the same organizational knowledge or language. Thus, although
participation in external information trading brings in new ideas resulting in
higher levels of individual creativity, the combination of external knowledge
with a firm’s existing knowledge may be time consuming, or may result in
solutions that are not tailored to the firm’s specific situation (Teigland, 2000).
In addition, information trading and reciprocity requires giving advice in
return. Helping others solve their problems can be quite time consuming, and
can lead to an individual devoting less time working on his or her company-
specific work tasks. Thus, an individual may have more difficulty in
completing his or her tasks on time and/or according to the task specifications
(Teigland, 2000). Thus, we have our next set of hypotheses:
HYPOTHESIS 7A: The higher the level of external information trading, the
higher the level of creativity.
HYPOTHESIS 7B: The higher the level of external information trading, the
lower the level of general performance.
Finally, for our last hypothesis, we look at the possible relationship

between internal and external information trading. In the stream of research by
Allen and colleagues, it was found that in many cases, individuals who had a
high degree of external communication activity also displayed a high degree of
internal communication activity (Allen & Cohen, 1969). These individuals
were labeled gatekeepers. Gatekeepers were found to display characteristics of
a buffering role, scanning and filtering information into the organization from
the outside world and then directing it as they felt necessary into the
organization. These individuals were often found to be in first-line supervisor
positions (Taylor, 1975). However, with access to the internet, it is now
possible for all individuals to become gatekeepers for their specialized
knowledge area, participating in external exchange and then recombining with
the specialized knowledge to the firm through internal integration. This brings
us to our final hypothesis:
HYPOTHESIS 8: The higher the level of external information trading, the

higher the level of internal information trading.
Figure 1 provides an overview of the hypotheses developed above.

376 ARTICLE FIVE
Figure 1 Model of Hypotheses
Co-located H1a(-),b
Coworkers
Non-co-located H2 Creativity
Coworkers
General
Electronic H3 H4a,b Performance
Networks
Internal
Information Trading
Contacts in H5 H8 H7a,b (-)

Other Firms External
H6 Information Trading
Extra-organizational
Electronic
Networks
RESEARCH SETTING
This research was undertaken in the Nordic operations (Denmark, Finland,
Norway, and Sweden) of Cap Gemini and was performed prior to the merger of
Cap Gemini and Ernst & Young Consulting. As a result, the company
description takes only the Cap Gemini organization into consideration. At the
time, Cap Gemini was Europe’s largest IT services and management consulting
company with more than 40 offices and 4,500 employees in the Nordic region
alone.
Within the Nordic region, Cap Gemini had numerous networks designed
to enhance the company’s knowledge management activities. We chose
participants in one electronic network, the NCN MS Community, because it
was recognized as a successful, vital conduit of knowledge exchange. This
electronic network had 345 members spread across the Nordic countries and
the members of this network all worked with applying Microsoft products in
their responsibilities with Cap Gemini. This particular population was chosen
for the study to ensure that research subjects had access to internal and external
sources of information and know-how, and had familiarity using the
communication technologies underlying information and know-how exchange
in electronic networks. In addition, the job responsibilities of the members of
the NCN MS Community required a considerable amount of creativity, as new
problem situations constantly arose due to the rapid pace of change in
information technology as well as the diversity among client project demands.
This helped ensure that the population chosen for this study had to balance both
general job performance and demands for creativity.
Study Design and Data Collection

In November 1999, we began the data collection through five in-depth
interviews with people involved in Cap Gemini Nordic’s knowledge
management operations. Interviews were conducted to better understand the
implementation and use of electronic discussion networks within Cap Gemini
and to design questionnaire items relevant to this specific organization. We
pilot-tested the questionnaire on a group of 15 programmers. Pilot test results
indicated that the survey instrument was too long, thus items outside the scope
of this research were dropped from the survey. The final survey instrument
was sent to each of the NCN MS Community members asking him or her to
complete the questionnaire during January 2000. Throughout the data
collection process, individuals were assured that their responses would be kept
confidential and that all results would be presented only on an aggregate level.
Of the initial 350 individuals, five emails were electronically returned
due to an invalid email address. We received a total of 83 usable survey
responses from the 345 participants with valid email addresses for a response
rate of 24%. The average age of the respondents was 35.6 years with an
average of 4.0 years employed at Cap Gemini and 7.7 years of experience in
their competence. The sample was 8% women. After consultation with Cap
Gemini management, it was found that the demographic characteristics of the
group of respondents were representative of those of the entire NCN MS
Community.
Measures
All variables were assessed through survey responses. Several different
approaches exist for measuring performance, including both self-reported and
third party measurements. However, following discussions with Cap Gemini’s
management, it became apparent that supervisor-rated or other performance
measures such as salary would be difficult to obtain due to issues of employee
confidentiality. Accordingly, we opted to measure performance via self-
reporting measures. Of interest is that a number of previous studies have found
self-reporting measures to be superior to third party measurements (Heneman,
1974; Wexley, Alexander, Greenawalt, & Couch, 1980) and not upwardly
biased (Churchill, Ford, Hartley, & Walker, 1985). As described earlier, two
measures of performance were measured: general performance and creativity.
The survey contained a series of 7-point Likert scale questions. The
dependent variables were measured by asking respondents to rate the extent of
their agreement on a 7-point Likert scale (1=strongly disagree, 4=agree,
7=strongly agree). The independent variables were assessed by asking
respondents to indicate how often they engage in specific knowledge activities
(1=several times a day, 2=once a day, 3=once every two days, 4=once a week,
378 ARTICLE FIVE
5=once every two weeks, 6=once a month, 7=more seldom). The independent
variables were then transformed to convert responses from an interval scale to a
ratio scale prior to analysis in order to conform to the ratio scaling conventions
of the dependent variables. Actual survey items are provided in table 1.
ANALYSIS AND RESULTS

We tested the hypotheses using partial least squares (PLS), and performed two
separate analyses independently for each dependent variable (creativity and
general performance). PLS can be used to analyze measurement and structural
models with multi-item constructs that include direct, indirect and interaction
effects, and has become widely used in IS research (Chin & Todd, 1995;
Compeau & Higgins, 1995; Gefen & Straub, 1997). PLS does not assume
multivariate normality among sample distributions and takes into account
measurement error when assessing the structural model (Wold, 1982). As a
result, it is particularly useful for analyzing constructs that include
measurement error and covariance. The results are interpreted in two stages –
measurement and structural. Because the combined analysis of the
measurement and structural models enables measurement errors to be included
as an integral part of the model, and factor analysis to be combined in one
operation with the hypotheses testing, the result is a more rigorous analysis of
the proposed research model (Bollen, 1989).
Unlike LISREL and EQS structural equation modeling techniques, PLS
breaks down models into segments, allowing researchers to work with small
sample sizes. When determining sample size, theorists suggests that a “rule of
thumb” for items is ten times the most complex construct’s number of
indicators or the largest number of paths leading to a latent construct. The most
complex construct in the measurement model has 4 indicators, and the largest
number of paths leading to a latent construct is 3. Thus with 83 respondents
(83/4, or 20.75 responses per indicator) the sample size is sufficient for
established PLS guidelines to proceed with analysis (Chin, 1998). The
theoretical model was estimated using PLS Graph 2.91 (Chin & Frye, 1996).
Measurement Model
A crucial step prior to testing the theoretical model is assessing the accuracy of
the measurement model. The goals of assessing the accuracy of the
measurement model are to demonstrate that the measures used are valid and
that they adequately reflect the underlying theoretical constructs. The first step
in PLS is to assess the convergent validity of the constructs of interest, by
examining the average variance extracted (AVE). The AVE attempts to
measure the amount of variance that a latent variable component captures from
its indicators relative to the amount due to measurement error. The AVE is
calculated by taking the sum of the squared component loadings to an indicator
and dividing by the sum of the squared component loadings plus the sum of the
error variance. It is recommended that the AVE should be greater than .50,
meaning that 50% or more variance of the indicators should be accounted for.
Individual survey items that make up a theoretical construct must also be
assessed for inter-item reliability. In PLS, the internal reliability and
consistency for a given block of indicators can be calculated using the internal
composite reliability (ICR) developed by Werts, Linn and Joreskog (1973).
The ICR is calculated by squaring the sum of component loadings to an
indicator, then dividing by the sum of squared loadings plus the sum of the
error terms. Interpreted like a Cronbach’s coefficient, acceptable values of an
ICR for perceptual measures should exceed .7 (Fornell & Larcker, 1981).
Values less than .7 imply that the items underlying the construct may be
unrelated, or may be measuring more than one construct.
Discriminant validity indicates the extent to which a given construct is
different from other constructs, the measures of the constructs are distinct, and
the indicators load on the appropriate construct (Messick, 1980). One criterion
for adequate discriminant validity is demonstrating that the construct shares
more variance with its measures than it shares with other constructs in the
model (Barclay, Higgins, & Thompson, 1995). One measure of discriminant
validity in PLS is the average variance explained (AVE). The AVE may be
compared with the shared variance among the latent variables (i.e. the square
root of the AVE should be greater than the correlation between a construct and
any other construct) (Chin, 1998). A second way to evaluate discriminant
validity is to examine the factor loadings of each indicator (Chin, 1998). Each
indicator should load higher on the construct of interest than on any other
factor.
Table 1 presents the factor loadings and cross-loadings for all indicators.
Each indicator loaded higher on its theoretical construct than on any other
factor, indicating discriminant validity. Table 2 presents descriptive statistics,
AVEs, ICRs, and correlations between constructs. All AVE values are greater
than the .5 cut-off point indicating adequate convergent validity. The square
root of the AVEs are presented on the diagonal of the correlation matrix, and
are greater than the corresponding correlations, indicating adequate
discriminant validity. All ICR values are greater that the .7 cut-off, indicating
adequate reliability.
Similar to findings in prior research, it appears that the knowledge
workers in this sample rely to a great extent on co-located coworkers as sources
of information. In addition, the survey respondents also reported a fairly high
amount of intra-organizational boundary spanning communications and internal
information trading. Finally, there are some indications that people engage in
boundary spanning activities using the same type of information sources
(interpersonal vs. electronic). For instance, there is a strong correlation
between the use of non-co-located coworkers as information sources and
external contacts. In addition, there is a correlation indicating that people who
access information from intra-organizational electronic networks also access
advice from inter-organizational networks.
380 ARTICLE FIVE
Table 1 Survey Items, Constructs, Item Loadings and Cross-Loadings
Construct and Survey Items

Knowledge Sources
How often do you use the below
information sources in your everyday 1 2 3 4 5 6 7 8 9
work?
1 Coworkers in my location 1.0 .28 .03 .01 -.13 .05 .04 -.17 -.13
2 Cap Gemini colleagues in another
.28 1.0 .55 .42 .20 .47 .21 .30 .13
location
3 Cap Gemini electronic communities .01 .48 .92 .37 .44 .35 -.01 .30 .22
NCN MS electronic communities .04 .53 .92 .35 .50 .35 -.03 .28 .26
4 Other contacts outside of Cap Gemini .01 .42 .39 1.0 .14 .35 .44 .31 .37
5 Internet: Discussion
-.13 .20 .51 .14 1.0 .40 .00 .24 .18
forums/electronic communities
Information Trading
How many times during the past year
have the following happened?
6 You were contacted by someone in
Cap Gemini for some specific .08 .38 .36 .27 .35 .89 .39 .53 .33
technical information
You gave out some specific technical
information to someone at Cap .11 .34 .36 .31 .39 .90 .40 .47 .31
Gemini
You sent formal, written
communications in the form of
.02 .44 .18 .27 .24 .72 .48 .43 .27
reports or data to someone inside of
Cap Gemini
You sent informal, written
-.03 .43 .37 .34 .34 .85 .41 .50 .28
reports or data to someone inside of
Cap Gemini
7 You were contacted by someone
outside of Cap Gemini for some -.10 .07 -.02 .36 .06 .47 .88 .36 .29
specific technical information
You gave out some specific technical
information to a person working -.02 .11 -.09 .32 .03 .40 .90 .24 .23
outside of Cap Gemini
You sent formal, written
.17 .28 -.04 .43 -.07 .42 .86 .27 .24
reports or data to someone outside of
Cap Gemini
You sent informal, written
.10 .26 .05 .45 .00 .47 .91 .36 .29
reports or data to someone outside of
Cap Gemini
Performance
8 My work tasks demand creative and
-.03 .36 .32 .22 .23 .38 .23 .76 .39
totally new ideas and solutions
The others at Cap Gemini think that I
-.13 .19 .14 .13 .03 .42 .31 .79 .39
am creative
My colleagues at Cap Gemini
-.21 .21 .30 .35 .30 .55 .30 .84 .63
consider me to be a guru
9 I have a high level of expertise in the
-.14 .08 .30 .31 .27 .39 .29 .59 .91
technology with which I work
Able to meet objectives set by your
immediate superior relative to your -.07 .19 .14 .35 .00 .18 .19 .39 .71
colleagues at Cap Gemini
Able to meet my deadlines relative to
-.03 .01 -.10 .08 -.14 .02 .12 .26 .60
your colleagues at Cap Gemini
Table 2 Descriptives, ICRs, Correlations and AVE Values
Std
Range Mean ICR 1 2 3 4 5 6 7 8 9
Dev
1 Co-located
0-3.7 3.0 1.0 n/a 1
Coworkers
2 Non Co-loc.
0-3.7 1.3 1.2 n/a .28 1
Coworkers
3 Intra-Org
0-3.4 1.2 1.1 .92 .03 .55 .92
Nets
4 Contacts at
0-3.7 .80 .87 n/a .01 .43 .39 1
other firms
5 Extra-Org -
0-3.7 1.3 1.3 n/a .20 .55 .14 1
Nets .13
6 Internal Info
0-3.1 1.2 .81 .91 .05 .47 .38 .36 .40 .84
Trading
7 External Info -
0-3.6 .56 .70 .94 .04 .21 .44 .01 .50 .89
Trading .02
8 -
Creativity 1.5-7 4.2 1.2 .84 .30 .32 .31 .24 .58 .35 .80
.17
9 General -
3-6.3 4.6 .88 .79 .13 .26 .37 .18 .35 .30 .61 .75
Performance .13
Structural Model
Table 3 summarizes the PLS structural analysis. Figure 2 provides a graphical
representation of the results. To evaluate the models, R2 values were calculated
for endogenous constructs. Interpreted like multiple regression results, the R2
indicates the amount of variance explained by the model (Chin, 1998). The
overall model explained 38% of the variance in creativity and 17% of the
variance in general performance. In addition, the model explained 44% of the
variance in internal information trading and 20% of the variance in external
information trading. Specifically, we find support for H1a, workers that rely on
co-located coworkers rate lower on creativity (b = -.20, p < .05). However,
382 ARTICLE FIVE
contrary to expectations in H1b, there was no relationship between accessing

information from coworkers and general performance (b = -.15, p ns). We
found support for H2, people who rely on coworkers in other locations are
more likely to engage in internal information trading (b = .24, p < .01), as well
as H3, people participating in internal electronic networks engage in internal
information trading (b = .26, p < .01). Finally, as predicted in hypotheses H4a
and H4b, internal information trading results in higher levels of creativity (b =
.54, p < .01) and higher levels of general performance (b = .28, p < .05).
Table 3 Results of PLS Analysis

Internal External
General
Info Info Creativity
Performance
Trading Trading
H1a,b Co-located Coworkers
-.20* -.15
H2 Non Co-located
Coworkers .24**
H3 Intra-organizational
Electronic Nets .26**
H4a,b Internal Information

Trading .54** .28*
H5 Contacts at Other Firms

.45**
H6 Extra-organizational
Electronic Nets -.06
H7a,b, External Information

H8 Trading .45** .09 .17
R2 .44 .20 .38 .17

* p < .05, two-tailed test
** p < .01, two-tailed test
As predicted in H5, people who use their contacts at other firms as

knowledge sources engage in external information trading (b = .45, p < .01).
However, we find no support for H6, thus there is no evidence for a
relationship between people participating in external electronic networks also
engaging in external information trading. Contrary to predictions, external
information trading did not directly influence creativity (H7a) or general
performance (H7b). Finally, we find support for H8, people who engage in
external information trading are also likely to engage in internal information
trading (b = .45, p < .01).
Figure 2 Model of Results
.23*
Co-located
Coworkers -.20*
R2=.38
Non-co-located
Coworkers Creativity
.24**
Intra-organizational General
Electronic 2 Performance
R =.44 .54**
Networks
Internal R2=.17
.26** Information Trading .28*
.45**
Contacts at .45**
Other Firms External
Information Trading
Extra-organizational
Electronic R2=.20
Networks

This research provides insight into how firms may create organizational
structures that balance efficient and flexible knowledge integration among
individuals. In terms of internal integration, individuals search for information
and know-how located both within and across intra-organizational boundaries
and integrate it with their own when performing work-related tasks. However,
our findings suggest that the view of the firm as a knowledge integrator needs
to be further developed by incorporating a dimension of cooperation. Grant’s
theory assumes that individuals are willing to share knowledge with each other
without expecting anything in return. Yet our research findings suggest that
knowledge integration through boundary spanning activities is supported by
informal information trading.
While knowledge can be accessed from co-located coworkers in one’s
communities of practice without expectation of reciprocal trading and returns,
once intra-organizational physical boundaries are crossed, expectations of
returns for knowledge sharing, as exhibited in information trading, appear to
come into play. In order for an individual to access knowledge from others
outside his or her immediate physical location, it appears that he or she must be
willing to give something in return. In addition, our results suggest insights
into the importance of flexible knowledge integration structures above and
beyond the efficiency of co-location.
384 ARTICLE FIVE
Efficiency and Flexibility of Integration: Implications for

Performance
In terms of integrative efficiency, we found evidence that a high reliance on co-
located coworkers results in lower levels of creativity. This suggests that the
knowledge of co-located coworkers and communities of practice may be
largely redundant and the integration of this local knowledge, although
efficient, may stifle the development of new ideas and innovations. On the
other hand, information that crosses intra-organizational boundaries enhances
integrative flexibility, as evidenced by the positive relationship between
internal information trading and both general performance and creativity.
Individuals may either reconfigure existing knowledge to fit their local needs or
integrate new know-how and innovative ideas with their own.
One surprising finding, though, is that integrative efficiency did not
enhance an individual’s general performance as predicted. One potential
explanation may be related to the task being performed. In software
development there is some degree of standardization and universal technical
language since individuals across the firm are typically using the same
underlying programming languages. Thus, individuals may trade standardized
programming hints or code across organizational boundaries that may simply
be “plug’n play”, enabling use with current know-how and easier
recombination with internal firm knowledge. Thus, in the area of software
development and the universal language of standardized programming code, it
may be just as efficient to integrate knowledge from co-located coworkers as
coworkers across the organization.
In terms of external integrative flexibility, contrary to expectations we
found that external information trading has no direct relationship to individual
performance, rather it affects creativity and general performance indirectly
through its influence on internal information trading. We had expected that
knowledge coming from contacts outside of the firm in one’s inter-
organizational networks of practice would be relatively novel and lead to more
creative solutions, but at the same time it would be more difficult to apply to
one’s task, thus requiring more time to use. One potential explanation for our
results may be that knowledge coming from outside the firm may be so novel
that it cannot be applied to any immediate solution. Rather its dissemination to
others and subsequent recombination with the firm’s knowledge is necessary to
adapt this knowledge to the firm’s specific use. Thus, the ability to develop
creative solutions and improve performance may involve combining existing
internal knowledge with novel external knowledge.
This finding supports theories of absorptive capacity, which suggest that
the firm’s ability to assimilate new, external information is largely a function of
the firm’s ability to internally process that information (Cohen & Levinthal,
1990). In addition, this research potentially offers additional insight: our
results suggest that performance is enhanced by the recombination of
knowledge that crosses internal organizational boundaries. This provides some
evidence that organizational absorptive capacity is enhanced through intra-
organizational cross-boundary knowledge flows, and by recombining
knowledge through informal information trading. While there has been a great
deal of emphasis placed on the importance of face-to-face interactions and co-
location for transferring valuable knowledge, organizations interested in
enhancing absorptive capacity may want to establish structures other than long-
term co-location that emphasize flexible knowledge integration.
Flexibility of Integration: Role of Electronic Discussion Networks

Our research findings also indicate that creating electronic social networks may
enhance integrative flexibility and information trading activities as well. In
terms of intra-organizational electronic networks, our findings provide support
for previous research suggesting that norms of reciprocity are critical for
sustaining knowledge exchange in electronic discussion networks (Lakhani &
von Hippel, 2000; Wasko & Faraj, 2000). However, contrary to expectations,
we find no relationship between participation in extra-organizational electronic
networks and external information trading. Thus, it seems that individuals are
more likely to engage in information trading with others with whom they have
a common bond, such as organizational membership.
One explanation may be found by looking at the relationship between
the ability to establish reciprocal exchange and various communication
channels. It may be more difficult to build reciprocal relationships with
individuals in extra-organizational electronic networks since members
generally have not met each other face-to-face and have little social influence
over one another due to the voluntary and anonymous nature of the exchange.
In addition, when reciprocity occurs in these networks it is typically of a
generic and not a dyadic nature (Kollock, 1999). In order for an individual to
give to the network, there must be a level of trust across the network members
that ensures other network members will “pay back” when requested.
Our findings indicate that it may be easier to build trust and achieve a
norm of reciprocity in intra-organizational electronic networks than in extra-
organizational ones. This may be because individuals within these networks
have a common organizational tie and are thus working for the greater good of
the company (Constant et al., 1996). However, there are other aspects to
consider. Intra-organizational networks may be more stable in terms of
participation, membership, and identification of participants. It is also possible
that individuals are not as anonymous as they are in extra-organizational
networks. Finally, misbehavior in an intra-organizational electronic network
may be more easily “punished” and carry tangible deterrents, while positive
behaviors may be rewarded through increases in status and reputation in the
organization. As a result, intra-organizational electronic networks may be
better able to control their boundaries and member behavior, resulting in more
effective flows of knowledge.
386 ARTICLE FIVE
Implications for Practice

These findings indicate that organizations concerned with knowledge
management may need to rethink their knowledge management strategies and
find a balance between reliance on co-located coworkers and the promotion of
flexible integration through boundary spanning communication to improve
individual performance and creativity. Thus, results from this study suggest an
important new use of internet-based communication technologies to support
knowledge management. Rather than using technology to replace traditional
knowledge management techniques, such as creating document repositories, we
need to think of non-traditional ways to leverage these new technologies for
improved knowledge flows within and across the firm, by leveraging networks
that support the exchange of advice and ideas between individuals.
While intra-organizational trading can be viewed positively without
question in terms of enhancing individual performance, and ultimately a firm’s
competitive advantage, the presence of inter-organizational information trading
draws into question the degree to which a company’s proprietary knowledge is
leaking across the firm’s boundaries. The decision to trade or not with external
parties is placed in the hands of an individual working for the firm. As such,
most economic and management researchers would argue that this informal
transfer is a disadvantage for the firm since the individual’s actions may not be
in line with the firm’s objectives, and may lead to a firm being unable to
capitalize on the unique benefits from an innovation (Schrader, 1991).
However, the results of this research seem to indicate that external
information trading is beneficial for the firm, although indirectly through
recombination with existing knowledge. Thus, while an individual may trade
away “proprietary” knowledge, the ability of a rival firm to turn this into an
innovation lies in its ability to internally integrate the new knowledge into the
existing knowledge base of the firm. In addition, trading information across
organizational boundaries enhances the inflow of new ideas and innovations.
Information trading also ensures that help will be reciprocated at a point where
the information seeker is in need of advice. Thus, while information trading
implies knowledge leakage across firm boundaries, it also ensures that new
knowledge flows back into the firm.
Limitations and Areas for Further Research

We should note the limitations of the study and caution that this study was of
an exploratory nature, with the contributions merely acting as guidelines for
further research. First, this study only examined knowledge workers focused
on developing software solutions in one company, thus limiting the
generalizability of our findings. Further research should examine individual
information trading across multiple organizations, and across multiple
categories of knowledge workers. In addition, the number of participants in
this study, although adequate for analysis, is relatively small to make
conclusive statements outside of this context. Further research should include
all organizational knowledge workers, the impact of location, and all internal as
well as external organizational information sources. Another limitation is our
use of self-reported survey measures only, increasing the risk of common-
method bias. Further research should include other performance data sources
in addition to survey data.
Conclusion
This paper examined the relationship between various knowledge integration
activities and individual knowledge worker performance. We found that
internal and external boundary spanning through participation in various
networks of practice facilitates information trading that in turn results in
improved individual performance. Therefore, creating organizational structures
that increase the flexibility of knowledge integration may support the creation
of new ideas and innovations, leading to sustainable competitive advantage. In
addition, this study indicates that people who rely on co-located coworkers as
information sources report lower levels of creativity. This indicates that
informal organizational structures such as communities of practice that enhance
the efficiency of knowledge integration, without regard to flexibility, may
“bind and blind” – supporting adherence to the same ideas and information,
potentially impeding the creation of new knowledge and stifling performance.
Therefore, organizations concerned with knowledge management and creative
solutions should focus on balancing knowledge integration structures that
support efficiency with flexibility, emphasizing boundary spanning and
informal information trading through both personal and electronic networks.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support provided by Cap Gemini
Nordic and in particular that of Christian Forsberg , Christian Storck, and Carl
Anlér as well as Victor Sylvan and Hanna Janson of the Stockholm School of
Economics.
REFERENCES
Alavi, M., & Leidner, D. 1999. Knowledge management systems: Issues,
challenges and benefits. Communication of the Association for Information
Systems, 1: 1-28.
Allen, T., & Cohen, S. 1969. Information flow in research and development
laboratories. Administrative Science Quarterly, 14: 12-19.
Allen, T. J. 1977. Managing the Flow of Technology. Cambridge, MA: MIT
Press.
Barclay, D., Higgins, C., & Thompson, R. 1995. The partial least squares (PLS)
approach to causal modeling: Personal computer adoption and use as an
illustration. Technology Studies, 2,2: 285-309.
Blau, P. M. 1964. Exchange and Power in Social Life. New York: Wiley.
388 ARTICLE FIVE
Bollen, K. A. 1989. Structural Equations with Latent Variables. New York:

John Wiley and Sons.
Bouty, I. 2000. Interpersonal and interaction influences on informal resource
Academy of Management Journal, 43,1: 50-66.
Brown, J. S., & Duguid, P. 1991. Organizational learning and communities-of-
practice: Toward a unified view of working, learning, and innovation.
Management Review, 40, 3: 90-111.
Brown, J.S. & Duguid, P. 2000. The Social Life of Information.
Carter, A. P. 1989. Knowhow trading as economic exchange. Research Policy,
18: 155-163.
Chin, W. W. 1998. The partial least squares approach to structural equation
modeling. In G. A. Marcoulides (ed.), Modern Methods for Business
Research. Mahway, New Jersey: Lawrence Erlbaum: 295-336.
Chin, W. W., & Frye, T. 1996. PLS Graph (Version 2.91.03.04). Calgary,
Canada: University of Calgary.
Chin, W. W., & Todd, P. A. 1995. On the use, usefulness, and ease of use of
structural equation modeling in MIS research: A note of caution. MIS
Quarterly, 19,2: 237-246.
Churchill, G. A., Ford, N. M., Hartley, S. W., & Walker, O. C. 1985. The
determinants of salesperson performance: A meta-analysis. Journal of
Marketing Research, 22: 103-118.
Cohen, W. M., & Levinthal, D. A. 1990. Absorptive capacity: A new
perspective on learning and innovation. Administrative Science Quarterly,
35: 128-152.
Compeau, D. R., & Higgins, C. A. 1995. Computer self-efficacy: Development
of a measure and initial test. Information Systems Research, 6,2: 118-143.
Science, 7,2: 119-135.
Davenport, T. H., & Prusak, L. 1998. Working Knowledge. Boston: Harvard
Faraj, S., & Wasko, M. M. 1998. The Web of Knowledge: An Investigation of
Self-Organizing
Communities of Practice on the Net. Paper presented at the INFORMS, Seattle,
Washington.
Fornell, C., & Larcker, D. 1981. Evaluating structural equation models with
unobservable variables and measurement error. Journal of Marketing
Research, 18: 39-50.
Fulk, J., & DeSanctis, G. 1995. Electronic communication and changing
organizational forms. Organization Science, 6,4: 337-350.
Gefen, D., & Straub, D. W. 1997. Gender differences in the perception and use
of e-mail: An extension of the technology acceptance model. MIS Quarterly,
21: 389-400.
Gerstberger, P., & Allen, T. J. 1968. Criteria used by research and development
Psychology, 52,4: 272-279.
Granovetter, M. 1983. The strength of weak ties: A network theory revisited.
Sociology, 91: 481-510.
7,4: 375-387.
Grant, R.M. 1996b. Toward a knowledge-based theory of the firm. Strategic
Management Journal, 17(Special Issue): 109-122.
Hauptman, O. 1986. Influence of task type on the relationship between
communication and performance: The case of software development. R & D
Management, 16,2: 127-139.
Heneman, H. G. 1974. Comparisons of self and superior ratings of managerial
performance. Journal of Applied Psychology, 59, 5: 638-642.
Hinds, P., & Kiesler, S. 1995. Communication across boundaries: Work,
Kogut, B., & Zander, U. 1992. Knowledge of the firm, combinative
capabilities, and the replication of technology. Organization Science, 3,3:
383-397.
Kollock, P. 1999. The economies of online cooperation: Gifts, and public
goods in cyberspace. In M. A. Smith & P. Kollock (eds.), Communities in
Cyberspace. London: Routledge: 220-239.
Lakhani, K., & von Hippel, E. 2000. How open source software works: "Free"
user-to-user assistance. Paper presented at the 3rd Intangibles Conference.
Business, NYU.
managing new product development. Strategic Management Journal,
13(Summer Special Issue): 111-125.
Levitt, B., & March, J. G. 1988. Organizational learning. Annual Review of
Sociology, 14: 319-340.
Messick, S. 1980. Test validity and the ethics of assessment. American
Psychologist, 35: 1012-1027.
Monge, P. R., Rothman, L. W., Eisenberg, E. M., Miller, K. I., & Kirste, K. K.
1985. The dynamics of organizational proximity. Management Science,
31,9: 1129-1141.
Nahapiet, J., & Ghoshal, S. 1998. Social capital, intellectual capital, and the
390 ARTICLE FIVE
Nohria, N., & Eccles, R. G. 1992. Face-to-face: Making network organizations

work. In N. Nohria & R. G. Eccles (eds.), Networks and Organizations:
Structure, Form and Action. Boston: Harvard Business School Press: 288-
308.
O'Reilly, C. 1982. Variations in decision makers' use of information sources:
The impact of quality and accessibility of information. Academy of
Management Journal, 25,4: 756-771.
Orlikowski, W. J. 1996. Learning from Notes: Organizational issues in
groupware implementation. In R. Kling (ed.), Computerization and
Controversy: Value Conflicts and Social Choices. New York: Academic
Press: 173-189.
Cornell University.
Pickering, J. M., & King, J. L. 1995. Hardwiring weak ties: Interorganizational
Spender, J.–C. 1996. Making knowledge the basis of a dynamic theory of the
firm. Strategic Management Journal, 17(Winter Special Issue): 45-62.
Sproull, L., & Faraj, S. 1995. Atheism, sex and databases: The net as a social
technology. In B. K. J. Keller (ed.), Public Access to the Internet.
Cambridge, MA: MIT Press: 62-81).
Szulanski, G. 1996. Exploring internal stickiness: Impediments to the transfer
of best practice within the firm. Strategic Management Journal, 17(Special
Issue): 27-43.
Taylor, R. L. 1975. The technological gatekeeper. R & D Management, 5: 239-
242.
Teigland, R. 2000. Communities of practice in a high technology firm. In J.
Birkinshaw & P. Hagström (eds.), The Flexible Firm: Capability
Management in Network Organizations. Oxford: Oxford University Press:
126-146.
York: John Wiley.
Science, 26,11: 1071-1085.
Van Maanen, J., & Barley, S. R. 1984. Occupational communities: Culture and
control in organizations. In B. M. Staw & L. L. Cummings (eds.), Research
in Organizational Behavior (Vol. 6.) Greenwich, CO: JAI Press.
Von Hippel, E. 1987. Cooperation between rivals: Informal know-how trading.
Von Hippel, E., & Schrader, S. 1996. "Managed" informal information trading:
Wasko, M., & Faraj, S. 1999. Webs of knowledge: An examination of
practice. Paper presented at the Academy of Management Conference,
Chicago, Ill.
Wasko, M., & Faraj, S. 2000. It is what one does: Why people participate and
Information Systems, 9,2-3: 155-173.
Wenger, E. 1998. Communities of Practice. Cambridge, UK: Cambridge
University Press.
Werts, C. E., Linn, R. L., & Joreskog, K. G. 1973. Intraclass reliability
estimates: Testing structural assumptions. Educational and Psychological
Measurement, 34: 25-33.
Wexley, K. N., Alexander, J. P., Greenawalt, J. P., & Couch, M. A. 1980.
Attitudinal congruence and similarity as related to interpersonal evaluation
in manager-subordinate dyads. Academy of Management Journal, 23, 2:
320-330.
Wold, H. 1982. Soft modeling: the basic design and some extensions. In H.
Wold (ed.), Systems Under Indirect Observation: Causality, Structure,
Prediction. New York: North Holland: pp. 1-47.
Article Six
Exploring the Relationships between Participation in

Networks of Practice, Centrality, and Individual
Performance in a Multinational Firm
In general, the terminology in this article corresponds to the terminology

in this thesis.
394 ARTICLE SIX
Previous version presented at INSNA - International Network of Social

Network Analysts Sunbelt Conference, 2002
Exploring the Relationships between Participation in

Networks of Practice, Centrality, and Individual
Performance in a Multinational Firm
Robin Teigland
Stockholm, Sweden
ABSTRACT
Multinational organizations create sustainable competitive advantage based on
their ability to effectively integrate knowledge that is increasingly dispersed
throughout their global operations. This knowledge resides in specialized form
among the organization’s individual members, and as individuals perform their
everyday work tasks, they participate in activities related to the firm’s
knowledge integration processes. In so doing, individuals build and participate
in emergent networks that have been labeled networks of practice, which
spread across the multinational’s intra-organizational boundaries as well as
across its legal boundaries. Grounded in the knowledge-based view of the firm
and in particular theories of knowledge integration, we investigate these
individual level activities and their relationship to individual outcomes of
centrality and performance. Using survey and social network data from a
multinational new media consulting company, we recreated the informal advice
networks for the entire multinational of 1698 individuals spread across 28
offices (84.7% response rate). We find results through structural equation
modeling that suggest that organizations should support individual level
activities that include not only the use of internal knowledge sources but also
the use of informal, external knowledge sources, such as participation in inter-
organizational networks of practice. Research results also suggest that there
are different patterns of activities related to knowledge integration depending
on whether efficient knowledge integration or flexible knowledge integration is
the goal. Implications for theories of the knowledge-based view of the firm,
the multinational, and networks of practice are discussed as well as some
implications for practice.
Keywords: knowledge, community of practice, network of practice,

multinational, social network analysis, structural equation modeling
NETWORKS OF PRACTICE IN A MULTINATIONAL 395
INTRODUCTION
To achieve competitive advantage, multinational organizations must
continuously create, transfer, and exploit knowledge that is increasingly
dispersed throughout their global operations (Bartlett & Ghoshal, 1989;
Hedlund & Nonaka, 1993; Doz & Hamel, 1997). Knowledge must be created
at a rapid pace while it is simultaneously transferred and applied throughout
these global operations. Coordination across subsidiaries prevents the
duplication of effort while at the same time ensuring the fastest time to market
with a product that customers want. Additionally, within the knowledge-based
view of the firm, it is argued that the challenge of a multinational is not to
divide a given task into activities to be performed efficiently by different
subsidiaries but to position the company so that “separate knowledge pieces”
from across the organization may be combined to initiate new tasks (Hedlund,
1994). The ability to create a sustainable competitive advantage is then based
on the firm’s combinative capability, or the ability to generate new applications
through the combination and recombination of existing knowledge (Kogut &
Zander, 1992). However, as many multinationals continue to expand their
operations and thereby increase the number of geographically dispersed
locations, employees, functions, and external partners, the task of effectively
making use of knowledge within the firm becomes more difficult. Both the
complexity of the multi-unit organizational structure and the differences in
language and local culture lead to significant challenges.
Recent research on multinationals is finding indications that
relationships of a more informal nature are playing an increasingly significant
role in the effective use of knowledge in these firms (Hansen, 1996, 1999; Tsai,
2002). In a review of the literature on coordinating mechanisms in
multinationals, Martinez & Jarillo (1989) find that since the mid-1970s
researchers have been paying considerably more attention to the importance of
“informal communication”72. Subsequent research has focused on knowledge
sharing through informal communication networks within multinationals and
has found a positive relationship between participation in informal intra-
organizational knowledge sharing and performance (Hansen, 1996). In these
studies, the level of analysis tends to be at the unit or project level with
researchers surveying unit managers about their subsidiary’s knowledge
sharing and social relations, e.g., socializing during events such as company
picnics (Tsai & Ghoshal, 1998). Yet there are few studies investigating the role
that the individual plays in the informal knowledge processes in multinationals,
despite proponents of the knowledge-based view of the firm arguing for the
72
Martinez & Jarillo (1989) define informal communication as communication that
occurs through informal networks, personal contacts, intra-subsidiary visits, meetings,
conferences and forums, and transfer of managers.
396 ARTICLE SIX
importance of the individual in these processes. For example, Grant proposes

that the firm be viewed as an enabler of coordinating action among individuals
while Hedlund suggests that one of the major characteristics that makes the
firm unique is its ability to transfer knowledge between levels of analysis - e.g.
from the individual level to the firm level and vice versa.
One body of literature that has been paying increasing attention to the
individual and his or her role in the knowledge processes of firms is that of
networks of practice (Brown & Duguid, 2000). Networks of practice are
groups of individuals connected together through social relationships that
emerge as individuals interact on task-related matters when conducting their
work. Researchers have investigated different forms of these networks of
practice in several settings, with communities of practice being the most well-
known and well-researched network of practice (Lave & Wenger, 1991;
Wenger, 1998). This literature argues that networks of practice are vital
conduits of knowledge within the firm and that they are the nexus of new
knowledge creation (Brown & Duguid, 1991). However, there are few
empirical studies focusing on performance issues related to networks of
practice or on networks of practice within multinationals. Thus, the purpose of
this chapter is to tie the above together by focusing on knowledge integration
through networks of practice within the setting of a multinational firm. In
particular, we investigate the relationship between an individual’s participation
in networks of practice, individual centrality, and an individual’s work-related
performance.
With the above in mind, we performed a unique study of a multinational
internet consulting firm, Icon Medialab (Icon). Through the administration of a
web-based questionnaire to all employees in the firm, we collected social
network data on the emergent work-related networks of all 1698 individuals
spread across 28 offices in 16 countries in Asia, the United States, and Europe.
With this data, we were able to build a rich picture of the firms’ emergent
work-related networks as well as the participation of the firm’s individuals in
networks reaching across the multinational’s boundaries. We then linked these
patterns to individual performance (creative vs. efficient) in order to provide
suggestive evidence of the value of knowledge flows in networks of practice.
Such inquiry makes three important contributions. First, this research
empirically examines the trade-offs between individuals accessing knowledge
from intra-organizational networks of practice located within their local
subsidiary or that span subsidiary boundaries as well as knowledge accessed
from inter-organizational networks of practice. Second, this research clarifies
how various patterns of individual level knowledge integration are related to
individual performance in complex knowledge environments. Finally, this
research makes possible more precise theoretical models of how multinational
organizations may design their organizations and their knowledge management
activities to support knowledge exchange and the creation of new knowledge to

enhance individual and thus organizational performance.
This article is organized as follows. In the following section, we briefly
review the relevant knowledge-based view of the firm and network of practice
literatures in addition to previous research on technology transfer within multi-
unit firms and social networks. These literatures provide the foundation for the
conceptual model and the specification of a set of hypotheses related to the
relationships between knowledge integration and individual performance.
Section three describes the research methodology and provides a description of
the research site. Section four reports the results of the empirical study while
the last section provides a discussion of the results and the implications of this
research for theory and practice.
Recent advances in strategic management thought suggest that organizational
resources and capabilities rather than served markets are the principal source of
sustainable competitive advantage and that knowledge is the most important
strategic resource of the firm. As a result, firms are increasingly being
described as distributed knowledge systems (Grant, 1996a,b; Spender, 1996;
Tsoukas, 1996). Assuming that knowledge is a critical input to production
processes, then organizational capability stems from the higher-ordered
organizing principles that structure relationships between individuals and the
various groups to which they belong in order to integrate the specialized
knowledge of individuals (Kogut & Zander, 1992; Zander & Kogut, 1995;
Grant, 1996a,b; Spender, 1996). This increased emphasis on organizational
capability and knowledge has led to the development of the knowledge-based
view of the firm (Kogut & Zander, 1992, Grant, 1996a,b; Spender, 1996).
Within this view, there is considerable emphasis on the individual. For
example, Nonaka (1994:17) states, “At a fundamental level, knowledge is
created by individuals. An organization cannot create knowledge without
individuals. The organization supports creative individuals or provides a
context for such individuals to create knowledge.” According to Grant
(1996a), competitive advantage results from how effective firms are in
integrating the specialized knowledge of their members. He further proposes
that this effectiveness depends upon the efficiency, the scope, and the
flexibility of knowledge integration. Efficiency refers to how productive firms
are in integrating individuals’ specialized knowledge. Efficient integration is
related to the frequency of interactions between individuals, where higher
levels of frequency engender automated responses from each organizational
member as well as a common language of discourse. The scope of knowledge
integration refers to the different types of specialized knowledge being
integrated – the more complex the scope, the greater the difficulty for
398 ARTICLE SIX
competitors to replicate. The flexibility of integration reflects extending

existing capabilities through boundary spanning activities in order to access
and reconfigure additional knowledge through both internal and external
integration.
Grant emphasizes the importance of integrative flexibility.
Hypercompetitive conditions in the marketplace drive the eventual erosion of
all positions of competitive advantage. Thus, sustaining a competitive
advantage requires flexibility and the creation of new capabilities. Firms need
to establish knowledge integration techniques that extend existing capabilities
by bringing in new knowledge and reconfiguring existing knowledge.
However, this need for flexibility to access new knowledge presents complex
organizational issues with regard to firm structure, firm boundaries, and the
choices between accessing knowledge from internal networks and external
networks.
Turning to multinational firms, one of the key issues underlying the
knowledge-based view in these settings is to understand how knowledge is
integrated across geographically dispersed units to create organizational
capability (Hansen, 1996). The literature on organizational coordination refers
to coordination as the integration or linking together of different parts of the
organization (Van de Ven et al., 1976; Tsai, 2002), and knowledge integration
within a multinational context has been defined as “the process of searching for
and transferring knowledge through the interunit network” (Hansen, 1996:1).
The extant multinational literature has identified two generic types of
coordinating or integrating mechanisms in multinationals: (1) formal
hierarchical structure and (2) informal lateral relations. As mentioned above,
recent research has increasingly emphasized the importance of these informal
relations (Martinez & Jarillo, 1986). Hansen (1996) finds in his research of a
high technology multinational that the vast majority of relationships between
company divisions were informal, with only 20% of the interdivisional contacts
more formalized, e.g., licensing agreements, regular technology meetings, etc.
Results further reveal that the more central an R&D team is in the
multinational’s informal network of units possessing the relevant expertise, the
easier the R&D team can conduct searches for the appropriate knowledge and
thus the faster the team’s completion time. A final finding is that if a unit has
weak relations within the multinational network, then the unit’s projects are
slowed down when the knowledge to be transferred is very complex.
We may, however, question these theories of knowledge integration and
multinationals on two premises. First, Grant’s theory of knowledge integration
represents a paradox: increasing the efficiency of integration may hinder
flexibility and the ability to create new innovations. For example, prior
research suggests that creating organizational structures that facilitate the
increase in efficiency of knowledge integration through common language and
frequent interactions may result in knowledge hoarding, less creativity and the
“not invented here syndrome” (Granovetter, 1973; Szulanski, 1996). Second,

proponents of knowledge integration seldom refer to issues of “cooperation”.
Grant’s theory focuses primarily on issues of coordination (i.e., structuring to
enhance the effectiveness of knowledge integration) while Hansen’s theory
views the central processes of knowledge integration within the multinational
as involving the search and transfer of knowledge through informal relations.
They both leave out a key component by assuming that people are willing to
share knowledge openly and freely if provided with the structures and
opportunities to interact. However, research by Tsai & Ghoshal (1998)
suggests that knowledge integration is dependent upon the degree to which
individuals develop trusting relationships in informal social interactions.
In addition, prior research on networks of practice that span a firm’s
legal boundaries suggests that individuals do not give away help and advice to
others in their informal social networks for free (von Hippel, 1987; Schrader,
1991, Macdonald & Williams, 1993; Bouty, 2000). Rather, individuals trade or
exchange knowledge with expectations of reciprocity. For example,
Macdonald & Williams (1993) find that individuals who are gatekeepers within
their organizations informally exchange knowledge with others outside their
organization in dyadic reciprocal relationships. In a more recent study, Isabelle
Bouty (2000) investigates the knowledge sharing decisions of researchers and
finds that individuals share their knowledge only with others with whom they
are mutually acquainted, share a high level of trust, and whom they do not
consider to be a competitor. Researchers find similar results when
investigating electronic networks of practice or groups of individuals who
communicate on work-related tasks primarily through computer-based
communication technologies, such as bulletin boards, listservs, etc (Lakhani &
von Hippel, 2000; Wasko & Faraj, 2000). For example, Wasko & Faraj (2000)
investigate participation in three technical usenet newsgroups and find that the
most frequent response to why respondents share their knowledge with others
is reciprocity. Additionally, the studies by von Hippel and Schrader revealed
that individuals participating in these inter-organizational reciprocal knowledge
exchanges often trade proprietary firm knowledge with one another.
In summary then, what is essential is that the expectation of reciprocity
is a key cooperation mechanism underlying knowledge exchange that crosses
organizational boundaries. However, few studies focusing on knowledge
integration within multinationals have investigated this issue of cooperation nor
have they investigated the underlying processes of knowledge exchange at the
individual level. Additionally, research into the relationship between
individual-level knowledge exchange and individual performance has been
extremely limited. As a step in this direction, we develop a set of hypotheses
relating individual participation in internal and external networks of practice,
knowledge exchange, and individual outcomes of centrality and performance.
400 ARTICLE SIX
RESEARCH MODEL AND HYPOTHESES

In this section, we look at the use of various sources to access knowledge
including participation in various networks of practice, internal and external
knowledge exchange, and the individual outcomes of centrality and individual
performance. We examine two types of individual performance: 1) efficient
performance or the ability to meet one’s job demands and 2) creative
performance or the ability to develop creative solutions. We expect that
participation in various networks of practice and the exchange activities
performed by individuals will impact their centrality and their general
performance and creative performance in different ways depending on whether
an individual’s activities are reflective of application of current knowledge or
concerned with new knowledge creation and innovation.
Efficiency of Integration
Internal Codified Sources. Within the knowledge-based view, competitive
advantage is partly dependent upon how efficiently the firm utilizes and
integrates its existing knowledge. As a result, firms develop systems for
capturing and making explicit the knowledge that is developed as individuals
conduct their tasks within the firm. Proponents of the knowledge-based view
build on Demsetz (1988) and argue that the firm is better able to do this than
the market since the firm provides the continuity of association between
individuals participating on the same task. As individuals work together over
time, they develop shared mental models and a common language that enable
them to codify their tacit knowledge. The firm can then store this codified
knowledge for reuse. Relative to individuals outside the firm, employees can
then easily access and reuse this codified knowledge as they share the same
communication code and mental models as those who codified it (Nonaka &
Takeuchi, 1995). Such codified storage is most visibly done using computer
databases and a company intranet that includes applications such as project
information repositories and skills databases as well as message boards. In
addition, due to the majority of work being performed and saved in digital
form, individuals have easy access to documents and other codified sources
that might not be found on the intranet. Thus, our argument is that at an
individual level we would expect to see an association between an individual’s
use of internal codified knowledge sources and their individual efficient
performance. In other words, an individual who is more inclined to search the
firm for already existing solutions and adapt them to his or her task, as opposed
to “reinventing the wheel” through developing one’s own solution, are more
likely to have a higher degree of efficient performance. However, a high reuse
of existing codified knowledge is unlikely to lead to the access of new
knowledge or the recombination of knowledge, which would affect creative
performance. Thus, we have the following two hypotheses:
HYPOTHESIS 1A: The higher the use of internal codified knowledge

sources (e.g., intranet, documents, electronic message boards) in task-
related matters, the higher the level of individual efficient performance.
HYPOTHESIS 1B: The higher the use of internal codified knowledge

sources (e.g., intranet, documents, electronic message boards) in task-
related matters, the lower the level of individual creative performance.
Co-located Coworkers. While organizations spend considerable resources on

documenting work tasks, research has consistently found that people prefer oral
sources to written sources (Allen, 1977). Codified knowledge sources are often
too strict to interpret each new situation and thus cannot fulfill the requirements
needed to perform the task (Brown & Duguid, 1991). In addition, Polanyi
(1962) describes knowledge as taking two forms, explicit and tacit. Explicit
knowledge is that which can be easily explained and codified, and tacit
knowledge is the additional knowledge that individuals are unable to articulate.
Tacit knowledge has a personal quality that makes it hard to formalize and
communicate and is deeply rooted in action, commitment and involvement in a
specific context (Polanyi, 1962). Some researchers argue that the most
effective means to transfer tacit knowledge may actually be not to codify it, but
to transfer it through an implicit mode (Yanow, 2000). According to Reber
(1993), transfer through an implicit mode means that “the acquisition of
knowledge takes place largely independently of conscious attempts to learn and
largely in the absence of explicit knowledge about what was acquired.” This
transfer of tacit knowledge thus requires transfer through word of mouth and
frequent interaction with others.
Ethnographic research on work practices finds that this frequent
interaction often occurs in communities of practice (Lave & Wenger, 1991;
Orr, 1996; Snyder, 1996; Wenger, 1998). Researchers have found that these
informal networks emerge over time between individuals working on similar
task-related issues located in face-to-face settings. With knowledge-intensive
tasks, often no one individual can solve the problem on his or her own due to
the inability to know everything. Thus, when an individual becomes stuck in
conducting a work-related task, he or she often turns to knowledge sources that
are the most easily accessed (such as asking co-located coworkers), rather than
searching for and using the best knowledge source (such as codified sources or
non-co-located coworkers) (Gerstberger & Allen, 1968; O'Reilly, 1982).
Through patterns of mutual exchange and collaboration, individuals share
knowledge to help each other reduce the equivocality of problematic issues and
build the community’s memory (Orr, 1996, Wenger, 1998). Thus, individuals
who rely on others in their local setting to a high degree are likely to engage in
a high degree of mutual knowledge exchange.
402 ARTICLE SIX
Through this mutual exchange and collaboration over time, individuals

become bound together by the context of the situation in an informal manner in
communities of practice (Brown & Duguid, 1991). These emergent structures
provide the nexus for the sharing and transfer of valuable individual and group
tacit knowledge (Kogut & Zander, 1992), resulting in higher performance of
the community as a whole (Brown & Duguid, 1991, 1998; Wenger, 1998).
Individuals develop a common language, explicit and tacit rules of behavior
and coordination, and a shared identity, (Wenger, 1998). Accessing knowledge
from others who share the same coding scheme and language is highly efficient
(Tushman & Katz, 1980), thus economizing on the amount and intensity of
communication needed to achieve knowledge integration. In addition, in many
work environments, employees are confronted with information overload.
Through asking someone in the community for help, time does not have to be
spent sorting though piles of information for relevant documents (Wenger,
1998). This suggests that one of the most efficient sources of knowledge
should be co-located coworkers, who are more likely to frequently interact with
each other and develop into a community of practice due to their sharing the
same physical space. Thus, due to the efficiency of integration, people who
access knowledge from co-located coworkers to a high degree should report
higher levels of individual efficient performance.
However, prior research argues that individuals within a social clique
tend to have strong ties, which have been defined as emotionally intense,
frequent, and involving multiple types of relationships, e.g., friends, advisors,
and coworkers (Granovetter, 1973). The result is that the knowledge held by
the members of a social clique tends to be redundant with that held by other
members, providing little additional information over what an individual may
already know (Granovetter, 1973, 1983). Thus, the knowledge available
through co-located coworkers is likely to be limited and superfluous, impeding
the ability to develop new and creative ideas. In addition, the highly efficient
structures that support knowledge integration and the exploitation of core
capabilities may evolve into core rigidities and competency traps –
inappropriate knowledge sets that preserve the status quo and limit new
insights, resulting in gaps between the knowledge of the firm and changing
market conditions (Levitt & March, 1988; Leonard-Barton, 1992). Therefore,
while accessing knowledge from co-located coworkers is likely to be highly
efficient and lead to better efficient performance, co-located coworkers are less
likely to offer the integrative flexibility needed to enhance creativity and
develop new capabilities. The above then leads to our next set of hypotheses:
HYPOTHESIS 2A: The greater the reliance on co-located coworkers as

sources of knowledge on task-related matters, the higher the level of
internal knowledge exchange.
HYPOTHESIS 2B: The greater the reliance on co-located coworkers as

sources of knowledge on task-related matters, the higher the level of
individual efficient performance.
HYPOTHESIS 2C:. The greater the reliance on co-located coworkers as

sources of knowledge on task-related matters, the lower the level of
individual creative performance.
Internal Flexibility of Integration

Non-co-located Coworkers. Internal flexibility of integration involves the
extent to which existing knowledge within the firm can be recombined and
reconfigured to create new knowledge (Grant, 1996). Knowledge that is
physically dispersed across a multinational may facilitate integrative flexibility
since it is more likely to be non-redundant to that which is found within the
same physical location (Granovetter, 1973). Advances in communication
technologies have made it easier for people who are working on similar task-
related problems yet physically dispersed across a multinational to
communicate, thus individuals may relatively easily access knowledge from
coworkers who are working in other offices. These new media greatly reduce
the cost of communicating with others and thus, the ability of individuals to
conduct knowledge integration through emergent networks across a
multinational has greatly increased. For example, email greatly increases the
possibility for individuals to access unknown people and new social circles
within multi-unit firms due to its asynchronous nature, the ability to
simultaneously communicate with more than one person at a time, as well as
the ability to easily forward messages (Feldman, 1987). As a result, the use of
email and other new media lead to new contacts that might not otherwise have
occurred, thus expanding the number and variety of people being used as
knowledge sources in problem-solving (Kraut & Attewell, 1993).
Previous research on occupational communities (e.g., van Maanen &
Barley, 1984) has shown that when people work in a similar occupation, e.g.,
software programmer, police, etc., they develop similar identities, values, and
vocabularies. This shared identity and language allow people to communicate,
regardless of whether they work in the same physical location or have a
previous history of a relationship. As such, individuals dispersed across an
organization working on similar tasks may create emergent networks through
which knowledge about practice can both travel rapidly and be assimilated
readily (Brown & Duguid, 2000). In this manner, intra-organizational
distributed networks of practice are similar to communities of practice in that a
shared practice is the substrate that ties members together. However, due to
less frequent patterns of interaction and lower intensities of social pressure,
non-co-located coworkers may be less willing or committed to exchange
knowledge without some type of return (Blau, 1964). People often prefer to
404 ARTICLE SIX
hoard their knowledge because they perceive that sharing knowledge results in
reduced status and personal worth (Orlikowski, 1996). This is especially so
when knowledge is the basis of a personal competitive advantage over others
(Thibaut & Kelley, 1959). Therefore, individuals are more likely to expect
reciprocity when engaging in knowledge exchange across internal
organizational boundaries, especially in situations where relations are not
characterized by frequent interactions and a high level of trust (Nahapiet &
Ghoshal, 1998). Thus, accessing advice and knowledge from non-co-located
coworkers is likely to result in internal information trading with the expectation
of reciprocity in return. This leads to our next hypothesis:
HYPOTHESIS 3: The greater the level of communication on task-related

matters with non-co-located coworkers to access knowledge, the higher
the level of internal knowledge exchange.
Integration Centrality. Individuals who are more highly central in

communities of practice in both local units as well as in the multinational’s
internal distributed networks of practice facilitate the firm’s ability to conduct
activities that promote internal integrative flexibility. Research on
communities of practice suggests that central individuals are influential in
shaping the flow of knowledge within the community as well as the future
direction of the development of the community’s knowledge since the more
highly an individual is embedded in a community of practice, the more others
turn to this individual for help and advice when solving problems (Schenkel et
al., 2002). In turn, we would expect that individuals central in a multinational’s
net of distributed networks of practice would also be influential in the efficient
and flexible knowledge integration processes of the firm since they influence
both the emergent knowledge flows between units and the knowledge
integration processes of other individuals. These individuals play the role of
brokers (Wenger, 1998) and are comparable to boundary spanners in the
technology transfer literature (Tushman & Scanlan, 1981). Through their
collaboration with others in distributed networks of practice, brokers gather
knowledge from areas across the organization and transfer it to their own
physical location while providing knowledge in reciprocation to other members
of their network of practice.
Research has shown members have different levels of community
participation; central individuals are full participants or “insiders” and are
highly embedded in the community while others are peripheral and less
embedded in the network of interdependent relations (Lave & Wenger, 1991).
Individuals who are “insiders” in communities of practice reach this position
through a process of legitimization during which the individual learns the
language and values of the community, while most importantly, how to
function as a community member (Lave & Wenger, 1991). For example, the
informal language of the practice can only be learned through informal

relationships since it is not the technical language of the trade such as that
taught in training manuals (Schenkel, 2002). However, an individual’s
community participation status is jointly determined by the community and the
individual. In other words, just because an individual wishes to become a
central individual does not necessarily guarantee the individual such
participation status. Rather, the individual must gain legitimacy within the
community. This process of legitimacy occurs only through mutual
engagement and knowledge exchange between the individual and other
community members over time (Lave & Wenger, 1991, Schenkel et al., 2002).
To date, comparable research on participation and membership status
within communities of practice has not yet been performed on intra-
organizational distributed networks of practice. However, we may hypothesize
that the dynamics are similar to communities of practice in that status is jointly
determined by the network and the individual. As such, the process of
legitimacy required to reach central status would only occur through mutual
engagement and knowledge exchange over time. As such, we now have our
next hypothesis:
HYPOTHESIS 4: The higher the level of internal knowledge exchange, the

higher the degree of integration centrality.
Creative Performance. While research on the relationship between centrality

and individual performance is rather limited. Previous research has provided
evidence of a link between centrality in a communication network and several
important variables that might lead to performance such as influence
(Burkhardt & Brass, 1990) and cognition (Walker, 1985) while more recent
research provides evidence of a direct positive relationship to individual
performance. In a study of business school alumni, Seibert, Kraimer, & Liden
(2001) find support for the role of access to information and resources as full
mediators of the relationship between social capital and career success.
Sparrowe, Liden, Wayne, & Kraimer (2001) find a positive relationship
between in-degree centrality (i.e., the number of relationships in which an
individual is sought out for advice) within one’s workgroup and individual
performance at five different organizations. Baldwin & Rice (1997)
demonstrate that centrality in the advice network of a sample of MBA students
is positively related to student grades. Finally, Mehra et al. (2001) show that
centrality within advice and workflow networks within a high-technology firm
predicts workplace performance while Ahuja et al. (2003) point out a direct
relationship between a central position in virtual R&D groups and individual
performance. In these studies, researchers generally find that it is the access to
unique or non-redundant knowledge that is an important factor in an
individual’s performance.
406 ARTICLE SIX
Research in networks of practice also suggests that those individuals

who are more central in the networks of practice within an organization also
have more opportunity to gain valuable and non-redundant knowledge from
others (Schenkel et al., 2002), and thus may exhibit a higher degree of
individual creative performance (Teigland & Wasko, 2003a,b). As mentioned
above, previous research by Hansen (1996) provides evidence that a central
position within a multinational facilitates the search for knowledge. Thus,
individuals who are central in a multinational’s distributed networks of practice
should be able to effectively access redundant knowledge across the
organization to help them conduct their own tasks. In addition, people who
engage in internal knowledge exchange are not only sending and receiving
task-specific knowledge, they are also helping each other by taking the time to
work through each other’s problems. When an individual works through
someone else’s problems, he or she often develops insights into new methods
and new applications for existing knowledge (Wenger, 1998). Exercising
intellect by helping others is also likely to help people maintain and even
improve their own technical skills as well as the ability to see new applications
of knowledge. In addition, individuals who help others are entitled to
reciprocity, gaining access to new ideas and innovations when needed. Thus,
we expect integration centrality to have a direct impact on individual creative
performance, leading to our next hypothesis:
HYPOTHESIS 5: The higher the degree of integration centrality, the higher

the level of individual creative performance.
External Flexibility of Integration

In addition to internal integrative flexibility, firms need to integrate new
knowledge found in the external environment to remain competitive (Cohen &
Levinthal, 1990). This knowledge may be accessed through either market or
relational contracts (Grant, 1996). While relational contracts tend to refer to
formal inter-organizational arrangements, these contracts also comprise
informal communication exchanges between individuals, and previous research
suggests that a high degree of a firm’s knowledge is imported through by a
firm’s members participating in emergent relationships that span a firm’s legal
boundaries (Macdonald, 1995).
With the rapid spread of the internet, the ability to participate in inter-
organizational networks of practice has greatly increased. As a result,
individuals throughout hierarchical levels and functional competence groups in
the organization are no longer limited to contacting organizational coworkers
or to searching within the company walls for knowledge or advice (Cronin &
Rosenbaum, 1994; Kettinger & Grover, 1997). Individuals may now just as
easily contact friends, ex-colleagues, or other acquaintances who work outside
the organization and even in rival firms (Sproull & Faraj, 1995; Faraj &
Wasko, 1998). Through participation in these networks, individuals may

integrate knowledge from within their organization with new ideas and
innovations accessed from outside their organization.
In addition to acquaintances, individuals may also quickly and
effortlessly access numerous networks of knowledgeable individuals with
whom they are not acquainted through means such as listservs, chat rooms,
discussion boards, etc. (Hinds & Kiesler, 1995; Constant et al., 1996). These
electronic networks of practice connect individuals sharing the same profession
but who are globally dispersed and typically strangers. These electronic
networks revolve around numerous technical (e.g., programming) and non-
technical (e.g., criminal law) issues, and they generally offer a much broader
source of expertise than at the individual’s own company due to the numerous
participants from many different backgrounds. In these electronic networks,
individuals are able to share information and know-how through mechanisms
that support posting and responding to questions, sharing stories of personal
experience, and discussing and debating issues relevant to the professional
community (Wasko & Faraj, 2000).
As mentioned above, prior research on inter-organizational networks of
practice demonstrates that reciprocity is one of the guiding principles in these
informal exchanges regardless of whether the other individuals are
acquaintances working in other firms or stranger in an electronic network of
practice (von Hippel, 1987; Schrader, 1991; Lakhani & von Hippel, 2000;
Wasko & Faraj, 2000). While previous research suggests that the growth of
inter-organizational internet-based communication is likely to be especially
rapid in organizations with a high degree of professionals (Pickering & King,
1995), there is no evidence to indicate that this activity is limited to
professionals. Thus, when individuals seek help with their work-related tasks,
they may easily contact individuals across the globe regardless of time and
their demographic characteristics, organizational setting, or local culture (Hinds
& Kiesler, 1995; Sproull & Faraj, 1995; Faraj & Wasko, 1998). As a result,
individuals may integrate knowledge relatively easily from within their
organization with new ideas and innovations accessed through communications
with individuals outside their organization. However, due to strong norms of
reciprocity within inter-organizational networks of practice, relying on contacts
in other organizations for advice obligates the knowledge-seeker to share
knowledge in return. Thus, accessing knowledge from external sources
requires reciprocation through external knowledge exchange and trading. This
leads to our next hypothesis:
HYPOTHESIS 6: The higher the use of informal external knowledge

sources for advice on task-related matters, the higher the level of external
knowledge exchange.
408 ARTICLE SIX
Knowledge Exchange. For our next hypothesis, we look at the possible

relationship between internal and external knowledge exchange. The
technology transfer literature finds that there is a significant degree of overlap
between communication star, boundary spanning, and gatekeeper activities.
Thus, individuals who are communication stars and boundary spanners are
more likely to be gatekeepers (Allen & Cohen, 1969; Tushman & Scanlan,
1981). As mentioned above, Macdonald & Williams (1993) provide further
evidence of this relationship since they find that individuals who are
gatekeepers within their organizations informally exchange knowledge with
others outside their organization in dyadic reciprocal relationships. The
implication of this is that individuals who conduct a high level of external
knowledge exchange are likely to conduct a high level of internal knowledge
exchange. While previous research also finds that gatekeepers are often in
first-line supervisor positions (Taylor, 1975), with access to the internet, all
individuals may now conduct external knowledge exchange within their
specialized knowledge area and then recombine this knowledge through
internal knowledge exchange. This then brings us to the following hypothesis:
HYPOTHESIS 7: The higher the level of external knowledge exchange, the

higher the level of internal knowledge exchange.
Human Capital. We expect that other factors, often referred to as human

capital, are associated with individual performance (both efficient and creative)
within the knowledge integration perspective. In this study we measure the
education level and general work experience. Thus, we have the following
hypotheses:
HYPOTHESIS 8A: The higher the level of education, the higher the level of
individual performance (both efficient and creative).
HYPOTHESIS 8B: The higher the level of experience, the higher the level of
individual performance (both efficient and creative).
Differences Based on Functional Task. Studies on networks of practice have

generally limited their observations to individuals conducting similar tasks.
However, we have the rather general proposition that the extent to which the
above approaches to knowledge integration affect individual performance will
be contingent on the nature of the task being performed. For example,
previous research has shown that task knowledge characterized by a more
universal nature such as software programming is easier to communicate across
a firm’s boundaries (Allen, Tushman & Lee, 1979). Space limitations prevent
a detailed discussion, but one would expect ceteris paribus that the more
universal and the more fast-changing the task, the more important it would be
for individuals to have ready access to external personal and codified sources
of knowledge. In terms of the specifics of this study, we have separated

individuals into three task groups on the basis that these groups rely on
knowledge acquisition from different sources to undertake their work
effectively: (1) Commercial and Support (CS): administration, sales,
management, project management, etc., (2) System and Software Group
(SSW): system architects, software programmers, etc., and (3) Design Group
(DG): human computer interface specialists, art directors, copy editors, etc.
However, it is difficult to speculate a priori to what degree and in which
manner the relationships hypothesized above will differ. Thus, we have our
last hypothesis:
HYPOTHESIS 9: Significant relationships will differ depending upon the

individual’s work-related tasks within the firm.
Figure 1 presents the fully developed research model with the appropriate
hypotheses labeled.
Figure 1 Model of Individual Activities Related to Knowledge Integration
Use of H3a
Internal Codified
H3b (-)
Sources
Efficient
H4b Performance
Communication
H4c (-)
with Co-located
Coworkers H10a,b
H4a
Education
Internal Experience
H5
Knowledge
Communication Exchange H7a
H6 H10a,b
with Non-co-located
Coworkers
H9 Integration H7b Creative
Centrality Performance
H8 External
Knowledge
Exchange
Use of External
Knowledge
Sources
410 ARTICLE SIX
METHODS
Sample and Procedures

We conducted this research in a single firm, Icon Medialab (Icon). The
investigation of only one site is common in network studies (see Marsden,
1990; Hansen, 1996) due to the requirements of a closed network when
studying individual relationships in social network analysis. The choice of
Icon was motivated primarily on the basis that it was a medium-sized
multinational that encompassed a wide variety of functional competences, e.g.,
system architecture, programming, management consulting, art direction,
human computer interface, etc., as well as 28 locations across Asia, the US, and
Europe. Additionally, one reason for choosing Icon is that its employees in all
functions are not only extremely adept at using new internet-based
communication media such as bulletin boards, chatrooms, email, etc. but they
also use these to a high degree in their everyday work. Finally, we chose Icon
since access to this firm was facilitated due to previous research by the author
at this firm (Teigland, 2000).
We conducted two phases of data collection and analysis. In the first
phase, thirty-five interviews were conducted throughout the firm to gain an
understanding of the various networks of practice within the firm as well as the
different inter-organizational ones in which Icon individuals participated.
Together with human computer interface specialists and a programmer, we then
constructed an extensive web-based social network survey in English since
English is the official company language. We pilot-tested the survey with one
individual across 15 different offices and across different technologies (e.g., PC
vs. Mac, Internet explorer vs. Netscape). We then made several changes to
avoid misinterpretations of the questions as well as to remove any technical
bugs in the survey.
In the second phase, we administered the web-based questionnaire
through the company’s intranet to all employees of Icon Medialab. Previous
research has suggested that electronic surveys using scale-type questions are no
less valid than paper surveys (Liefeld, 1988). It has also been found that some
subjects prefer electronic surveys to paper (Newsted, 1985), and that email
responses may even be more valid (Sproull & Kiesler, 1986). It is important to
note here that all employees had access to their own computer and the internet
since the majority of their work was performed using the computer. In
addition, management placed few constraints on employees regarding the
internal or external use of any form of computer-mediated channels.
We placed a hyperlink to the survey at the top of the homepage on the
company’s intranet such that individuals could easily find the survey. In
addition, there were hyperlinks to the survey within the introductory email
from the researcher as well as within all reminder emails. Due to the length of
the survey, we designed the survey such that when the individual moved from
one survey section to the next, the individual’s answers were automatically
saved in the survey database. In this manner, an individual could leave the
survey and return at any time through the intranet link to find his or her
previously entered answers. To administer the survey, several mailings were
sent out by email to each individual, including 1) an initial request for
participation from the CEO of each office, 2) a request from the researcher with
a link to the survey, 3) a follow-up two weeks after the first mailing, 4) and if
necessary, a second follow-up three weeks after the first mailing.
Because our research required the complete network, we had to specify
a boundary around it. We used the membership criterion (Marsden, 1990;
Wasserman & Faust, 1994:31), thus we included those individuals who were
formally part of the organization. Individuals who were currently on leave of
absence, working only part-time, or were independent consultants working for
the company were eliminated from the respondent pool since their networks
would not be comparable to those employees who were actively working full-
time for the organization. The resulting number of total potential respondents
was 1698. We received 1439 completed surveys for a response rate of 84.7%,
a level considered to be high enough to perform sociometric network analyses.
Throughout the data collection process, individuals were assured that their
responses would be kept confidential on a secure server at the company’s third
party intranet host and that results would only presented in aggregate form. To
encourage responses, we entered all respondents into a drawing for 14 prizes of
approximately US $1600 in total value.
The average age of the respondents was 30.5 years (s.d. 5.82) with an
average of 590 (s.d., 409) days employed at Icon. Individuals had worked an
average of 3.03 (s.d. 1.02) years in their competence and 73.5% had the
equivalent of a university degree or higher. The sample was 39.6% women. In
terms of the split between the three task groups, (1) the Commercial and
Support Group had together 697 individuals (450 respondents in commercially
oriented functions, e.g., project managers, sales personnel, management
consultants, and 247 in support functions, e.g., finance, legal, human resources,
etc., (2) 454 in the System and Software Group, e.g., programmers, system
architects, etc., and (3) 288 in the Design Group, e.g., web designers, art
directors, human computer interaction specialists, etc.. After comparing our
sample with the entire Icon population, we find that the demographic
characteristics of the group of respondents were representative of those of the
entire multinational.
412 ARTICLE SIX
Survey Measures
Knowledge Sources and Knowledge Exchange

The survey contained a series of 7-point Likert scale questions that investigated
the use of various internal and external knowledge sources as well as
knowledge exchange activity. These independent variables were assessed by
asking respondents to indicate how often they used specific knowledge sources
(1=few times a day, 2=once a day, 3=few times a week, 4=once a week, 5=few
times a month, 6=once a month, 7=less than once a month). The internal
knowledge sources constructs included 1) use of internal codified sources (3
items), 2) communication with co-located coworkers (2 items), and 3)
communication with non-co-located coworkers (2 items). The external
knowledge sources constructs included informal external sources, e.g.,
electronic networks of practice, friends, colleagues (3 items). Both the internal
and external knowledge sources scales were adapted from Teigland (2000) and
Teigland & Wasko (2003a,b). The internal exchange scale comprised three
items and the external knowledge scale comprised four items. Both scales were
adapted from Leifer & Huber (1977) and Teigland & Wasko (2003b). The
frequency independent variables were then transformed to convert responses
from an interval scale to a ratio scale.
Integration Centrality
While the recall of brief, episodic interactions is highly inaccurate (Bernard,
Killworth, Kronenfeld, & Sailer, 1984), people are remarkably able to
accurately remember typical interactions and long-term relationships with other
individuals (Freeman, Romney, & Freeman, 1987), which are important for our
study. Thus, in order to determine integration centrality, we assessed relations
by asking respondents two questions: 1) “In general, which persons inside Icon
do you contact for help or advice when you are not sure what to do with your
work, i.e., for help or advice related to your tasks and not your administrative
activities?” and 2) “In general, who contacts you in the same way?” The lists
of individuals on the survey were directly linked to the company’s employee
database so that it would automatically always be current. However, we found
that there was a much larger number (n=2200) of individuals listed in the
company’s database than the number of individuals who were relevant for this
study. This was because the company’s database included individuals who
were on leave of absence, were independent consultants, had quit but not yet
been removed, or who were going to be quitting shortly. After several
iterations and pilot-testing, we decided to create one web page for each
individual office that listed the names of all the individuals within each office
alphabetically by first name and not last name since pilot-testing revealed that
individuals could recall first names and office to a much better degree than last
names. Next to each individual’s name and function were eight radio buttons,
four indicating the degree with which the respondent contacted the individual
listed and four indicating the degree to which the individual contacted the
respondent (1-4 scale indicating daily, weekly, monthly, or less than monthly
communication).
We placed a drop-down menu with an alphabetical listing of all the
offices within the organization at the top of the screen. In addition, we placed
buttons with “Next Office” and “Previous Office” at the top and bottom of the
screen. In this manner, respondents could easily move between offices,
locating others with whom they had relationships outside of their own office.
In order to ensure that people listed others outside of their own location, we
wrote the following, “Please think of people in ALL ICON OFFICES, not just
those in your own Icon office. To go to another office, click on Next Office or
choose another office from the drop-down menu.”
Using this network data, we constructed a measure of integration
centrality for each individual. Before making any calculations, we went
through the network matrices and removed all individuals who were not active,
full-time employees, as well as checked and corrected all individual
background data (e.g., office, title, competence, hierarchical level, etc.). We
then based our measure of integration centrality on two frequently used social
network measures: in-degree centrality and closeness centrality. Degree
centrality is calculated by simply counting the number of links to (in-degree)
and from (out-degree) an actor and this measure was used in the technology
transfer studies described above. However, unlike the above studies, our study
considers only the in-degree relations in the network for the degree centrality
measure. In other words, for each focal individual, we counted only those links
that other individuals reported that they turned to the focal individual for advice
(Sparrowe et al., 2001). In this manner, we determined the degree to which
each individual was sought out by others for advice and knowledge. It is
important to point out that unlike out-degree, in-degree centrality does not
suffer from the limitations of self-reports thus we were able to avoid the
potential problems of common method bias with this measure.
In addition to in-degree centrality, we calculated the degree of closeness
centrality (Freeman, 1979) for each individual in the firm. Closeness centrality
denotes the degree to which an individual is embedded in a network, i.e., how
close he or she is to all other individuals within the network, either directly
(e.g., a friend) or indirectly (e.g., a friend of a friend, a friend of a friend of a
friend, etc.). This measure is calculated by summing the lengths of the shortest
paths from one actor to all other actors in the network. It takes into account
both direct and indirect links by counting direct links as one step while giving
indirect links proportionally less weight. Thus, an individual who is maximally
close to all others in the network would have direct, unmediated relationships
with all others in the network while individuals who have indirect relations to
others have lower levels of closeness depending upon the number of
414 ARTICLE SIX
intervening nodes between him or her and all other individuals in the network
(Baldwin & Rice, 1997). In terms of knowledge networks, actors with a higher
level of closeness have greater and faster access to the knowledge of all other
individuals throughout the firm than those with lower levels of closeness.
Since we were interested in the individual’s general embeddedness in
the firm’s networks of practice and not the direction of his or her relationships
(i.e., whether the individual goes to another or vice versa), we used data from
both the advice network questions in the closeness measure. First we
transposed the second matrix that asked “In general, who contacts you in the
same way?” in order to make the two advice matrices equal in terms of
direction between the individual respondents. We then pooled the two matrices
using the average method. The calculation of closeness centrality requires
dichotomized (1 or 0) and symmetrical (i.e., non-directional) relations. Thus,
our next step was to dichotomize the data by converting all values to either a
“1” or “0” with a cutoff point at all values greater than 0.5 in the pooled matrix.
In this manner, we removed all values that had an original input as “4” (less
than monthly contact recoded as 1 in the network matrix) in the “I contact”
section that was not reciprocated by the alter respondent in the “Contacts me”
section, i.e., (0+1)/2=0.5. Our final step was to symmetrize the data using the
maximum rule; 98% of the pairs were symmetric.
Prior research has shown that there is a high degree of overlap between
the two measures of centrality that we used, i.e., individuals who have a high
level of in-degree centrality are also likely to have a high level of closeness
centrality. However, in a firm with many units spread across the globe, there is
the possibility that the two might not coincide to the degree that would be
found within a single unit firm. For example, an individual may have a high in-
degree within his or her own local unit. However, he or she might not have a
high level of closeness within the entire organization if his or her local
community of practice is not well connected with the rest of the organization.
On the same token, an individual who is well embedded in the firm due to a
few relationships with individuals who are in turn highly embedded may not
necessarily be one to whom a considerable number of others turn for advice.
Thus, by combining in-degree centrality with closeness centrality in one
measure, we arrived at a construct that we feel provides an adequate description
of an individual’s integration centrality within a multinational firm. In this
manner, we also depart from traditional social network analysis, which
generally only uses one measure at a time. We calculated in-degree and
closeness centrality scores for each individual using UCINET V (Borgatti,
Everett, & Freeman, 1992).
Human Capital
For the human capital variables, respondents were asked to indicate their
highest obtained educational degree (1=high school, 2=technical certificate,
3=bachelor, 4=master, 5=Ph.D.) as well as their work experience in terms of

the number of years they had worked in their competence or a similar one.
Individual Performance Variables

Individual performance is one of the most central and fundamental constructs
of organizational behavior; however, measuring it has proven to be a difficult
task. While there exist several different approaches, for example subjective
measures (e.g., self, peer, and supervisor ratings) and objective measures based
on direct measures of countable behaviors or outcomes (e.g., total sales
volumes or sales commissions for salespeople), the correlations between the
various measures tend to be less than “perfect” (see Bommer, Johnson, Rich,
Podsakoff, & MacKenzie (1995) and Harris & Schaubroeck (1988) for a
discussion). Researchers have also come to some agreement that perfectly
reliable and valid third party performance ratings are unattainable since they
are subject to a variety of biases, such as external conditions, the experience of
the rater with the job being evaluated, or the ability of the rater to observe the
ratee (Borman, 1978; Weekley & Gier, 1989). Thus, there exists no one “best”
measure of individual performance.
In our discussions with management regarding to what extent
performance measures were possible to collect, it became apparent that
supervisor ratings, peer ratings, or other performance measures such as salary
would be difficult to obtain due to issues such as employee confidentiality.
Accordingly, we opted to measure individual performance via self-reporting
measures. However, previous research at Icon supports this choice of self-
reported measures since we found that that supervisors and individuals at Icon
were in considerable agreement over the degree of individual performance.
Thus, for the purposes of this study we used two different subjective
dependent variables that measure individual performance: creative performance
and efficient performance. As discussed above, these measures represent the
two dimensions of efficiency of integration and flexibility of integration where
efficiency is manifested as the ability to meet deadlines and objectives and
flexibility is manifested as the ability to develop and implement new ideas,
processes, routines. Interestingly, while it may be somewhat difficult to
distinguish between these two measures, we do feel that it is important to
measure both since it is often difficult to develop solutions that are highly
innovative but that also meet objectives and deadlines.
Efficient Performance. Individuals were asked to answer three questions on

their ability to meet objectives and deadlines based on a seven-point scale
(1=extremely below average to 7= extremely above average) that created an
efficiency scale. This scale was adapted from Teigland (2000) and Teigland &
Wasko (2003a).
416 ARTICLE SIX
Creative Performance. Individuals were asked to answer four questions that

created a creative performance scale. This scale was taken from a larger
individual performance measure that has been used considerably to measure
innovative performance (Welbourne et al., 1998). The questions were based on
a seven-point scale (1=strongly disagree to 7=strongly agree). These items
represent creative performance since they incorporate not only the development
of new ideas, routines, and processes, but also their implementation.
ANALYSES AND RESULTS
Analyses
We used structural equation modeling to analyze the data for the entire sample
as well as for the three task groups: (1) Commercial and Support, (2) System
and Software Group, and (3) Design Group. Our first step was to conduct two
analyses in order to investigate any possible effects of method variance:
principal component analysis and confirmatory factor analysis. First, we
subjected all scale items to a principal component analysis using varimax
rotation. From this analysis, the expected factors clearly emerged. In addition,
the highest cross-loading of any one indicator on another factor was .203.
Table 1 provides the means, standard deviations, and correlations of the
variables. Second, we used confirmatory factor analysis and created a single
factor model in which all our measures loaded onto one factor, a method
variance factor. This single-factor model fit the data very poorly, which is
described below (Turban & Dougherty, 1994).
Table 1 Means, Standard Deviations, and Correlationsa

Variables Mean s.d. 1 2 3 4 5 6 7 8 9 10
1.Efficient Performance 5.09 0.84
2.Creative Performance 5.44 0.82 .36
3.Internal Codified Sources 2.38 1.39 .05 .07
4.Comm. w/ Co-located 4.11 0.95 .07 .07 .33
5.Comm. w/ Non-co-located 1.39 1.42 .07 .13 .38 .24
6.External Sources 2.74 1.18 .07 .11 .28 .19 .24
7.Internal Knowledge
Exchange 3.19 1.34 .09 .16 .31 .38 .29 .11
8.External Knowledge
Exchange 1.79 1.35 .09 .14 .17 .10 .21 .29 .38
9.Integration Centrality 18.88 9.16 .06 .07 .21 .23 .29 .10 .21 .05
10.Education 2.95 1.04 .08 .10 .01 -.04 .00 -.08 .05 .05 .04
11.Experience 3.03 1.20 .14 .21 .00 -.02 .16 .07 .03 .14 .10 .09
a
N = 1439. Correlations greater than or equal to .06 are significant at p < .05.
We then examined the hypothesized relationships among the use of

knowledge sources, knowledge exchange, integration centrality, and individual
performance. The use of structural equation modeling facilitates the
simultaneous examination of the relationships and an assessment of the fit of

the hypothesized model to the data as well as a test of the individual hypotheses
(Turban & Dougherty, 1994). We assessed model fit using several statistics. It
is widely accepted that reliance on the chi-square test alone is not
recommended due to its sensitivity to sample size, i.e., models that fit the data
reasonably well are often rejected due to medium to large sample size in which
the test is conducted (Bentler, 1980; Bentler & Bonett, 1980). Thus, we
conducted three additional fit tests that are not sensitive to sample size – (1) the
non-normed fit index (NNFI, Bentler & Bonett, 1980), (2) the comparative fit
index (CFI, Bentler, 1990), and (3) the root mean square error approximation
(RMSEA, Steiger, 1990). The CFI is the most highly recommended fit index
(Bagozzi & Edwards, 1998) and this fit indicates the relative improvement of
the fit of the hypothesized model over the null model, in which all observed
variables are specified as uncorrelated. When the hypothesized model is true in
the population, these indices have an expected value of 1.00 (Ashford,
Rothbard, Piderit, & Dutton, 1998). A value of .90 or higher is suggested to
indicate adequate fit (Bentler& Bonett, 1980). The RMSEA is also a widely
used fit statistic, providing an estimate of the discrepancy between the original
and the reproduced covariance matrices in the population. RMSEA values of
.05 are suggested to represent a close fit while values of less than .08 represent
a reasonable fit (Cudeck & Brown, 1983; Ashford et al., 1998).
For the entire sample as well as the three subsets, we estimated several
models and compared them to the null model as various authors have discussed
previously (Turban & Dougherty, 1994). More specifically we estimated the
(1) null model that was used as a baseline model, (2) an uncorrelated-latent-
variables model in which we loaded the manifest variables onto the latent
constructs and no paths were indicated between the latent variables, (3) the
theoretical model presented in figure 1, and (4) the modified theoretical model
that provided the best fit for the different samples. For the fourth model, we
attempted to find the best fit by both dropping paths as well as relaxing some of
the assumptions of the theoretical model by adding direct paths between the
independent variables and the dependent variables.
Results for Entire Sample

For the entire sample, the complete model with all the hypothesized
relationships has a chi-square of 1917 with 358 degrees of freedom, an NNFI
of .89, a CFI of .90, and an RMSEA of .055. Due to the RMSEA being over
.05, we decided to attempt to achieve a better fit. After dropping a few paths as
well as relaxing a few relationships, we obtained a model with a better fit. This
best fit model has a chi-square of 1567 with 359 degrees of freedom, an NNFI
of .91, a CFI of .92, and an RMSEA of .048. We therefore retained this
modified model as the best fitting model. Table 2 provides an overview of the
fits of the different models for the entire sample.
418 ARTICLE SIX
Table 2 Overview of SEM Results for Entire Sample

Model Ȥ 2 (df) ǻ Ȥ2 (ǻ df) RMSEA CFI NNFI
Null 15 897 (406) -- .140 .31 .26
Uncorrelated 3 247 (379) 12 650 (27) .073 .82 .80
Theoretical 1 917 (358) 1 330 (21) .055 .90 .89
Best Fit 1 557 (357) 360 (1) .048 .92 .91
We then examined the standardized parameter estimates to determine whether

the hypothesized relationships were significant and in the predicted directions.
Figure 2 and table 3 provide an overview of the results for the entire sample.
Table 3 Overview of Results from SEM for Entire Sample

Entire
Sample
Hypothesized Relationships
H1a: Use of Internal Codified sources – Efficient Performance
H1b: Use of Internal Codified Sources – Creative Performance
H2a: Comm. with Co-loc. Coworkers – Internal Exchange .46***
H2b: Comm. with Co-loc. Coworkers – Efficient Performance .13***
H2c: Comm. with Co-loc. Coworkers – Creative Performance .13***
H3: Comm. with Non-co-loc. coworkers – Internal Exchange
H4: Internal Exchange – Integration Centrality .14***
H5: Integration Centrality – Creative Performance .11***
H6: Use of External Knowledge Sources – External Exchange .34***
H7: External Exchange – Internal Exchange .27***
Human Capital
H8a: Education – Efficient Performance .06*
H8a: Education – Creative Performance .08*
H8b: Experience – Efficient Performance .14***
H8b: Experience – Creative Performance .20***
Non-hypothesized Relationships
Use of Internal Codified sources – External Exchange .12***
Comm. with Non-co-loc. Coworkers – Integration Centrality .60***
Use of External Knowledge Sources – Integration Centrality -.11***
Model Fit
Ȥ2 1557
Df 357
CFI .92
NNFI .91
RMSEA .048
* p < .05
** p < .01
*** p < .001
Efficiency of Integration. Specifically, Hypotheses 1a and 1b, relating the use

of internal codified sources to individual performance, receive no support.
However, we receive support for Hypothesis 2a relating communication with
co-located coworkers positively to internal knowledge exchange since we find
a statistically significant parameter estimate for this relationship (b = .46, p <
.001). Hypotheses 2b and 2c relate communication with co-located coworkers
positively to efficient performance (2b) and negatively to creative performance
(2c). We find a statistically significant parameter estimate for efficient
performance (2b; b = .13, p < .001) and creative performance (2c; b = .13, p <
.001). However, the parameter estimate for creative performance is in the
opposite direction than predicted – a positive instead of a negative relationship.
Internal Flexibility of Integration. We find a statistically significant parameter

estimate for the path between communication with non-co-located coworkers
and internal knowledge exchange (b = .09, p < .01), indicating support for
Hypothesis 3. Hypothesis 4 is supported since we find a statistically significant
parameter estimate for the relationship between internal knowledge exchange
and integration centrality (b = .14, p < .001). Hypothesis 5 relates integration
centrality to individual creative performance, and we find support for this
hypothesis (b = .11, p < .001), indicating that individuals with a higher level of
integration centrality have higher reported levels of creative performance.
External Flexibility of Integration. Respondents who use external knowledge

sources to a high degree report higher levels of external knowledge exchange,
providing support for Hypothesis 6 (b = .34, p < .001). Individuals who report
higher levels of external knowledge exchange also report higher levels of
internal knowledge exchange, providing support for Hypothesis 7 (b = .27, p <
.001).
Human Capital. Finally, in support of Hypotheses 8a and 8b, we find

statistically significant parameter estimates for the relationships between
education and experience with individual performance – education to efficient
performance (H8a; b = .06, p < .05), education to creative performance (H8a; b
= .08, p < .05), experience to efficient performance (H8b; b = .14, p < .001),
and experience to creative performance (H8b; b = .20, p < .001).
420 ARTICLE SIX
Best Fit. In addition, although not hypothesized, we find three other significant
path estimates. First, a significant parameter estimate is found for the
relationship between communication with non-co-located coworkers and
integration centrality (b = .60, p < .001). Second, we find a negatively
significant parameter estimate between external knowledge sources and
integration centrality (b = -.11, p < .001). Third, results provide a statistically
significant parameter estimate for the path between internal codified sources
and external knowledge exchange (b = .12, p < .001).
Results for Commercial and Support Group

For the Commercial and Support Group (CSG), the complete model with all the
hypothesized relationships has a chi-square of 1021 with 358 degrees of
freedom, an NNFI of .88, a CFI of .89, and an RMSEA of .054. Since all the
fit measures were not above the accepted rate for an adequate fit, we decided to
attempt to achieve a better fit. After dropping a few paths as well as relaxing a
few relationships, we achieved a model with a chi-square of 902 with 334
degrees of freedom, an NNFI of .90, a CFI of .92, and an RMSEA of .049. We
therefore retained this modified model as the best fitting model. Table 4
provides an overview of the fits of the different models for the Commercial and
Support Group.
Figure 2 Results of Structural Equation Model for Entire Sample – Best

Fita
Internal
codified Efficient
sources .12*** performance
.13*** .06* .14***

Communication
with co-located Education Experience
coworkers .46***
.13*** .08* .20***

Internal
knowledge Flexible
exchange performance
.14***
.09**
Communication .27***
with non-co-located .60*** Integration .11***
coworkers centrality
External
knowledge
exchange -.12***
.34***
External
knowledge
sources
a
Fit: Ȥ 2 = 1557, df = 357; NNFI = .91, CFI = .92, RMSEA = .048. Only statistically
significant paths are shown. Hypothesized relationships are represented by bold arrows, and
relationships that were not hypothesized are represented by light arrows. Hypothesized paths
that were not significant were eliminated from the model.
Table 4 Overview of SEM Results for CSG

Null 7 165 (406) -- .131 .33 .28
Uncorrelated 1 674 (379) 5 491 (27) .069 .84 .83
Theoretical 1 091 (358) 583 (21) .054 .89 .88
Best fit 902 (334) 189 (24) .049 .92 .90
Looking at the standardized parameter estimates to determine which
hypothesized relationships were significant and in the predicted directions, we
find a different set of results than for the entire sample. Figure 3 and table 5
provide an overview of the results.
Table 5 Overview of Results from SEM for CSG

Commercial
and Support
Group
H1a: Use of Internal Codified Sources – Efficient Performance
H2a: Comm. with Co-loc. Coworkers – Internal Exchange .28**
H2b: Comm. with Co-loc. Coworkers – Efficient Performance
H2c: Comm. with Co-loc. Coworkers – Creative Performance
H3: Comm. with Non-co-loc. Coworkers – Internal Exchange
H4: Internal Exchange – Integration Centrality .14**
H5: Integration Centrality – Creative Performance .15***
Human Capital
H8a: Education – Efficient Performance
H8a: Education – Creative Performance
H8b: Experience – Efficient Performance .13**
Use of Internal Codified Sources – Internal Exchange .14***
Comm. with Non-co-loc. Coworkers – Integration Centrality .48***
Use of External Knowledge Sources – Integration Centrality -.20***
Model Fit
Ȥ2 902
Df 334
CFI .92
NNFI .90
RMSEA .049
* p < .05
** p < .01
*** p < .001
422 ARTICLE SIX
Efficiency of Integration. Specifically, Hypotheses 1a and 1b, relating internal

codified sources to individual performance, receive no support. Hypothesis 2a
positively relates communication with co-located coworkers to internal
knowledge exchange, and a statistically significant parameter estimate is found
for this relationship (b = .28, p < .001). Hypotheses 2b and 2c relate
communication with co-located coworkers positively to efficient performance
(2b) and negatively to creative performance (2c). However, no support is
found for these relationships.
Internal and External Flexibility of Integration. No statistically significant

parameter estimate is found for the path between communication with non-co-
located coworkers and internal knowledge exchange, providing no support for
Hypothesis 3. Hypothesis 4 is supported since we find a statistically significant
parameter estimate for the relationship between internal knowledge exchange
and integration centrality (b = .14, p < .001). Hypothesis 5 relates integration
centrality to individual creative performance, and we find support for this
hypothesis (b = .15, p < .001), indicating that individuals with a higher level of
integration centrality have higher reported levels of creative performance. We
find support for the relationship between external knowledge sources and
external knowledge exchange, Hypothesis 6 (b = .40, p < .001). Respondents
who report higher levels of external knowledge exchange also report higher
levels of internal knowledge exchange, providing support for Hypothesis 7 (b =
.33, p < .001).
Human Capital. Finally, for Hypotheses 8a and 8b, we find statistically

significant parameter estimates for the relationships between experience and
individual performance – experience to efficient performance (8b; b = .13, p <
.01), and experience to creative performance (8b; b = .17, p < .001). However,
we find no significant relationship between education and either efficient or
creative performance
Best Fit. As for the relationships that were not hypothesized, we find three
other significant path estimates. First, the path between internal codified
sources and internal knowledge exchange is statistically significant (b = .14, p
< .01). Thus, those individuals in the periphery who use internal codified
sources to a higher degree also report higher levels of internal knowledge
exchange. Second, a significant parameter estimate is found for the
relationship between communication with non-co-located coworkers and
integration centrality (b = .48, p < .001). Third, we find a negatively
significant parameter estimate between external knowledge sources and
integration centrality (b = -.20, p < .001).
Figure 3 Results of SEM for Commercial and Support -Best Fita
Use of
Internal Codified Efficient
Sources Performance
.14**
.13**
.11**
Communication
with Co-located Education Experience
Coworkers .28**
.17***
Internal
Knowledge Creative
Exchange .14** Performance
.15***
with Non-co-located .48*** Integration
Coworkers Centrality
External
Knowledge -.20***
Exchange
.40***
Use of External
Knowledge
Sources
a
Fit: Ȥ 2 = 902, df = 334; NNFI = .90, CFI = .92, RMSEA = .049. Only statistically significant
paths are shown. Hypothesized relationships are represented by bold arrows, and relationships
that were not hypothesized are represented by light arrows. Hypothesized paths that were not
significant were eliminated from the model.
Results for System and Software Group

For the System and Software Group (SSW), the complete model with all the
hypothesized relationships has a chi-square of 821 with 358 degrees of
freedom, an NNFI of .90, a CFI of .91, and an RMSEA of .054. Since the
RMSEA is over .05, we decided to attempt to achieve a better fit. After
dropping a few paths as well as relaxing a few relationships, we achieved a
model with a chi-square of 745 with 335 degrees of freedom, an NNFI of .91, a
CFI of .92, and an RMSEA of .052. While the RMSEA is over 0.5, the other
measures indicate a marginally adequate fit. We therefore retained this
modified model as the best fitting model. Table 6 provides an overview of the
fits of the different models for the System and Software Group.
424 ARTICLE SIX
Table 6 Overview of SEM Results for System and Software Group

Null 5417 (406) -- .148 0.26 0.20
Uncorrelated 1192 (379) 4 225 (27) .069 0.84 0.83
Theoretical 821 (358) 371 (21) .054 0.91 0.90
Best fit 745 (335) 76 (23) .052 0.92 0.91
find yet another set of results. Figure 4 and table 7 provide an overview of the
results.
Table 7 Overview of Results from SEM for SSW

SSW
Group
H1b: Use of Internal Codified Sources – Creative Performance -.13*
H2b: Comm. with Co-loc. Coworkers – Efficient Performance
H2c: Comm. with Co-loc. Coworkers – Creative Performance
H3: Comm. with Non-co-loc. Coworkers – Internal Exchange .13*
H4: Internal Exchange – Integration Centrality .37***
H5: Integration Centrality – Creative Performance
Human Capital
H8b: Experience – Efficient Performance .12*
Use of External Knowledge Sources – Integration Centrality .26**
Model Fit
Ȥ2 745
df 335
CFI .92
NNFI .91
RMSEA .052
* p < .05
** p < .01
*** p < .001
Efficiency of Integration. Specifically, Hypothesis 1a, relating internal

codified sources to efficient performance, is refuted. A significant path relating
internal codified sources to creative performance is found. However, the
direction is a negative one and not a positive one as hypothesized (b = .13, p <
.05). Thus, individuals in the System and Software Group who use internal
codified sources to a higher degree report significantly lower levels of creative
performance. Hypothesis 2a positively relates communication with co-located
coworkers to internal knowledge exchange, and a statistically significant
parameter estimate is found for this relationship (b = .58, p < .001).
Hypotheses 2b and 2c relate communication with co-located coworkers
(2c). However, no support is found for either of these relationships.
Internal and External Flexibility of Integration. Support is found for

Hypothesis 3 since a statistically significant parameter estimate is found for the
path between communication with non-co-located coworkers and internal
knowledge exchange (b = .13, p < .05). Hypothesis 4 is supported since we
find a statistically significant parameter estimate for the relationship between
internal knowledge exchange and integration centrality (b = .37, p < .001).
Hypothesis 5 relates integration centrality to individual creative performance,
but the results failed to support this path. We find support for Hypothesis 6, the
relationship between external knowledge sources and external knowledge
exchange (b = .18, p < .001). Respondents who report higher levels of external
knowledge exchange also report higher levels of internal knowledge exchange,
providing support for Hypothesis 7 (b = .19, p < .001).

Best Fit. As for the relationships that were not hypothesized, we find one other
significant path estimate. A significant parameter estimate is found for the
relationship between external knowledge sources and integration centrality (b =
.26, p < .01). Thus, individuals in the System and Software Group who use
external knowledge sources to a higher degree are found to have higher levels
of integration centrality.
426 ARTICLE SIX
Figure 4 Results of SEM for System and Software Group – Best Fita
Use of
Sources Performance
-.13*
.12*
Communication
Coworkers .58***
.24***
Internal
Knowledge Creative
Exchange .37*** Performance
.13*
with Non-co-located Integration
External
Knowledge
Exchange .26**
.18***
External
Knowledge
Sources
a
that were not hypothesized are 2represented by light arrows. Hypothesized paths that were not
significant were eliminated from the model
Results for Design Group

For the Design Group, the complete model with all the hypothesized
relationships has a chi-square of 611 with 358 degrees of freedom, an NNFI of
.90, a CFI of .91, and an RMSEA of .050. Since the model fit measures could
be considered to be marginal, we decided to attempt to achieve a better fit.
After dropping a few paths as well as relaxing a few relationships, we achieved
a model with a chi-square of 546 with 335 degrees of freedom, an NNFI of .92,
a CFI of .92, and an RMSEA of .047. We therefore retained this modified
model as the best fitting model. Table 8 provides an overview of the fits of the
different models for the Design Group.
Table 8 Overview of SEM Results for Design Group

Null 3215 (406) -- .136 .24 .29
Uncorrelated 811 (379) 2 404 (27) .060 .85 .84
Theoretical 611 (358) 200 (21) .050 .91 .90
Best fit 546 (335) 65 (23) .047 .92 .92

find a fourth set of results. Figure 8 and table 9 provide an overview of the
results.
Table 9 Overview of Results from SEM for Design Group

Design
Group
H2b: Comm. with Co-loc. Coworkers – Efficient Performance .20**
H2c: Comm. with Co-loc. Coworkers – Creative Performance .23**
H3: Comm. with Non-co-loc. Coworkers – Internal Exchange
H4: Internal Exchange – Integration Centrality
H5: Integration Centrality – Creative Performance
Human Capital
H8b: Experience – Efficient Performance .18**
H8b: Experience – Creative Performance .19**
Use of Internal Codified Sources – Internal Exchange
Use of Internal Codified Sources – External Exchange
Use of Internal Codified Sources – Integration Centrality .23**
Comm. with Non-co-loc. Coworkers – Integration Centrality
Internal Exchange – Creative Performance .19*
Use of External Knowledge Sources – Integration Centrality
Model Fit
Ȥ2 546
Df 335
CFI .92
NNFI .92
RMSEA .047
* p < .05
** p < .01
*** p < .001
428 ARTICLE SIX
Efficiency of Integration. Specifically, the results fail to provide support for

Hypotheses 1a and 1b, relating internal codified sources to individual
performance. Hypothesis 2a positively relates communication with co-located
coworkers to internal knowledge exchange, and a statistically significant
parameter estimate is found for this relationship (b = .36, p < .001).
Hypotheses 2b and 2c relate communication with co-located coworkers
(2c), and this time statistically significant parameter estimates for both of these
hypothesized relationships are found. A positive relationship is found between
communication with co-located coworkers and efficient performance (b = .20,
p < .01). The path between communication with co-located coworkers and
creative performance is also statistically significant. However, it is in the
opposite direction than predicted (b = .23, p < .01). Thus, individuals who
communicate to a higher degree with their co-located coworkers also report
higher levels of creative performance.
Internal and External Flexibility of Integration. No support is found for

Hypothesis 3, the path between communication with non-co-located coworkers
and internal knowledge exchange, nor for Hypothesis 4, the path between
internal knowledge exchange and integration centrality. In the same vein,
results fail to support Hypothesis 5, relating integration centrality to individual
creative performance. Hypothesis 6, the relationship between external
knowledge sources and external knowledge exchange, receives support (b =
.32, p < .001). Respondents who report higher levels of external knowledge
exchange also report higher levels of internal knowledge exchange, providing
support for Hypothesis 7 (b = .37, p < .001).

Best Fit. As for the relationships that were not hypothesized, we find two other
significant path estimates. First, a significant parameter estimate is found for
the relationship between internal codified sources and integration centrality (b
= .33, p < .001). Thus, individuals in the Design Group who use internal
codified sources to a higher degree are also found to have higher levels of
integration centrality. Second, a statistically significant relationship is found
between internal knowledge exchange and creative performance (b = .19, p <
.05).
Figure 8 Results of SEM on Design Group - Best Fita
Use of
Sources .33*** .20** Performance
.18**
Communication
Coworkers
.36*** .19**
.23**
Internal
.19*
Knowledge Creative
Exchange performance
with Non-co-located Integration
External
Knowledge
Exchange
.32***
Use of External
Knowledge
Sources
a
that were not hypothesized are represented by light arrows. Hypothesized paths that were not
significant were eliminated from the model.
Results for the Entire Sample

An overview of the significant hypothesized relationships as well as those that
were not hypothesized is presented in table 10. While we do not find any
support for the hypotheses relating to the use of internal codified sources, we
do find strong or moderate support for several of the hypotheses involving the
use of internal interpersonal communication sources, the use of external
knowledge sources, as well as the two human capital variables of education and
experience. What is also interesting is the fact that we see such different results
for each dependent performance variable as well as for each of the three task
groups.
430 ARTICLE SIX
Table 10 Overview of Results from Structural Equation Modeling

Entire Comm.
Sample &Supp. SSW Design
H1a: Use of Internal Codified Sources – Efficient
Performance
H1b: Use of Internal Codified Sources – Creative -.13*
Performance
H2a: Comm. with Co-loc. Coworkers – Internal .46*** .28** .58*** .36***
Exchange
H2b: Comm. with Co-loc. Coworkers – Efficient .13*** .20**
Performance
H2c: Comm. with Co-loc. Coworkers – Creative .13*** .23**
Performance
H3: Comm. with Non-co-loc. Coworkers – .13*
Internal Exchange
H4: Internal Exchange – Integration Centrality .14*** .14** .37***
H5: Integration Centrality – Creative Performance .11*** .15***
H6: Use of External Knowledge Sources – .34*** .40*** .18*** .32***
External Exchange
H7: External Exchange – Internal Exchange .27*** .33*** .19*** .37***
Human Capital
H8a: Education – Efficient Performance .06*
H8a: Education – Creative Performance .08*
H8b: Experience – Efficient Performance .14*** .13** .12* .18**
H8b: Experience – Creative Performance .20*** .17*** .24*** .19**
Use of Internal Codified Sources – Internal .14***
Exchange
Use of Internal Codified Sources – External .12***
Exchange
Use of Internal Codified Sources – Integration .23**
Centrality
Comm. with Non-co-loc. Coworkers–Integration .60*** .48***
Centrality
Internal Exchange – Creative Performance .19*
Use of External Knowledge Sources –Integration -.11*** -.20*** .26**
Centrality
Model Fit
Ȥ2 1557 902 745 546
Df 357 334 335 335
CFI .92 .92 .92 .92
NNFI .91 .90 .91 .92
RMSEA .048 .049 .052 .047
* p < .05 ** p < .01 *** p < .001
Efficiency of Integration. Looking at internal codified sources, we do not find

any significant relationship between the use of these and performance for the
sample as a whole. Previous research at Icon has shown that when choosing to
use a codified source, some individuals at Icon prefer to go outside the firm,
and that they find external sources to be more helpful than internal ones
(Teigland, 2000). This is despite management having spent considerable
resources in building a company intranet. One reason may be due to the quick
pace of development of the industry within which Icon is located. Employees
may feel that the internal knowledge is “old”, thus it could not help with “new”
problems. As interviews in the pilot study revealed, knowledge and solutions
developed for use in projects could already be out of date within six months of
their development, thus rendering them useless. However, we would argue that
there is still a lot of process knowledge such as project management knowledge
that could be developed and codified for further reuse internally. This is
perhaps reflected in the relatively higher use of internal codified sources by
individuals in the Commercial and Support Group than individuals in either the
SSW or Design Groups.
If we then look at interpersonal communication, we see the level of
communication with co-located coworkers as a predictor of efficient
performance. This is in keeping with our a priori expectations. A high degree
of internal personal interaction with other members of one’s communities of
practice who share the same language should be a highly efficient source of
knowledge. However, we did not expect to find a positive relationship between
communication with co-located coworkers and creative performance. Based on
previous research (Teigland & Wasko, 2003a), we had predicted a negative
relationship since it is expected that the knowledge of an individual’s local
coworkers is largely redundant, thus hampering the creation of new ideas
through flexible knowledge integration. One explanation for our present
finding may be due to the difference between the two studies. In Teigland &
Wasko (2003a), the group of co-located coworkers is characterized by one
functional discipline, software programming. Thus, when one individual
discusses with other co-located individuals, they tend to share the same
functional competence. However, in the Icon study, due to the organization of
the company as described above, the group of co-located individuals comprises
several functional disciplines. Interviews also revealed that communities of
practice could contain a variety of functions as well. In our study, we measure
communication with co-located coworkers as based on communication both
with those in the same competence as well as with those in another
competence. Thus, the more an individual communicates with co-located
coworkers, the more likely that these other individuals may be from several
disciplines and that his or her communities of practice comprise multiple
disciplines. Since flexible integration involves new patterns of integrating
existing knowledge, the ability to achieve this through combining different
432 ARTICLE SIX
bodies of functional knowledge is increased the more an individual

communicates with individuals from a variety of functional areas, which then
may impact creative performance.
A second explanation may lie in the definition of creative performance.
As discussed above, flexible integration is not only developing new ideas and
processes through the recombination of knowledge, but it also entails putting
these new ideas and processes into action. A good new idea is only worth
something to the firm if results can be achieved from its implementation. The
implementation of new ideas and processes may be dependent upon the ability
of an individual to understand how to recombine knowledge for application in a
local context, which an individual may only develop through a high level of
interaction with co-located coworkers. Previous research by the author merely
looked at creativity or the development of new ideas without considering
whether these new ideas were implemented (Teigland & Wasko, 2003a), while
this research takes into account both angles as is called for by the definition of
the flexibility of integration.
Internal Flexibility of Integration. Additionally, we find support for the

relationship between integration centrality and creative performance. Thus, our
results suggest that efficient performance is dependent upon the ability of an
individual to reuse existing local knowledge without having to tap into the
firm’s global knowledge base. However, a higher level of creative
performance is partially dependent upon an individual’s position in the firm,
indicating that flexible knowledge integration is the result of the recombination
of knowledge found throughout the firm’s geographically dispersed locations.
In terms of the drivers of integration centrality, we do find the expected
positive relationships between internal knowledge exchange and integration
centrality. For co-located individuals, participation in the mutual engagement
of sharing and receiving knowledge leads to integration centrality in the firm as
a whole. This supports community of practice theory that argues that
movement into a central position within a community is dependent upon
mutual engagement between the individual and others in the community and
that this movement is jointly determined.
However, we also find a direct relationship between communication
with non-co-located coworkers and integration centrality. This finding is
unexpected since we argued that internal knowledge exchange would fully
mediate this relationship. Our findings indicate that a central position in the
multinational’s networks of practice is only partially dependent upon an
individual’s participation in exchange relationships of sharing and receiving
knowledge with others both locally as well as in other locations. In this study,
interactions with non-co-located coworkers are not dependent upon exchange
relationships to the same degree as interactions with co-located coworkers.
Once local organizational boundaries are crossed and interactions occur
primarily through electronic communication channels, mutual engagement

expressed in terms of knowledge exchange between geographically dispersed
individuals is no longer as important in predicting integration centrality. Thus,
we may conclude that for this multinational, norms of mutual engagement are
not as strong in non-face-to-face relationships as for relationships based upon
face-to-face relationships as a determinant of integration centrality. This
finding then suggests that communities of practice and distributed networks of
practice may exhibit different norms of behavior.
External Flexibility of Integration. Turning to external integrative flexibility

and the use of external knowledge sources, our results support our hypotheses.
The use of external knowledge sources does not have a direct impact on an
individual’s performance. Rather it affects creative performance indirectly
through its influence on internal knowledge exchange and integration
centrality. In addition to the hypothesized relationships, we also find a direct
negative relationship between the use of external knowledge sources and
centrality. Thus, those individuals who merely use external knowledge sources
to a high degree but do not integrate this knowledge with that of others through
internal knowledge exchange remain on the outskirts of the firm and by
implication achieve lower levels of creative performance. These findings
support our previous research (Teigland & Wasko, 2003b). In new-media
companies such as Icon, employees are able to communicate across external
organizational boundaries with others working on similar problems and as a
result, access new knowledge and ideas. Knowledge coming from outside the
firm may be so novel that it cannot be applied to any immediate solution
without being placed within the context of the firm. Application is then
facilitated through a high degree of personal interaction with co-located as well
as non-co-located coworkers. Knowledge is disseminated to others as well as
recombined with the firm’s knowledge in order to adapt this external
knowledge to the firm’s specific use. Thus, the ability to develop and
implement new solutions and improve performance may involve combining
existing internal knowledge with novel external knowledge.
Results for the Three Task Groups

In addition to these findings, what also stand out are the differences in patterns
for the three group of individuals based on work-related tasks. One of the main
differences across these groups is the use of internal codified sources and the
relationship of the use of these with knowledge exchange, integration
centrality, and performance. These underlying patterns may also partially
explain the reason as to why we find no relationship to internal codified sources
for the sample as a whole.
434 ARTICLE SIX
Commercial and Support Group

If we look first at the Commercial and Support Group, one primary difference
is that there is no direct relationship between communication with co-located
coworkers and either creative or efficient performance as predicted and found
for the sample as a whole. The relationship to efficient performance is
mediated by internal knowledge exchange and the relationship to creative
performance is mediated through internal knowledge exchange and integration
centrality. In addition, the use of internal codified knowledge sources is of
considerable importance for this group. Internal codified sources have a direct
positive relationship to internal knowledge exchange while internal knowledge
exchange has a direct positive relationship to efficient performance. For this
group, efficient solutions could best be achieved by reusing existing solutions
that were accessed through the firm’s intranet and documents as opposed to
“reinventing the wheel” through developing one’s own solution. Exchanging
knowledge with one’s local coworkers makes it easier to find appropriate
solutions.
However, the Commercial and Support Group also exhibits a different
pattern for creative performance. Creative performance is solely dependent
upon an individual’s integration centrality, which is in turn predicted by
communication with non-co-located coworkers as well as internal knowledge
trading with one’s co-located coworkers. In addition, the use of informal
external knowledge sources also plays a significant role in integration
centrality. Individuals who only use external knowledge sources do not move
into central positions due to the negative relationship between the use of these
external sources and integration centrality. However, individuals who
exchange knowledge to a high degree both internally and externally are central
individuals. Thus, new solutions and processes are created and implemented
through the combination of local internal knowledge with external knowledge
and may be facilitated by the access to timely strategic knowledge found
through communication with non-co-located coworkers.
Design Group
Looking at the Design Group, the use of internal codified sources is again
important for this group, as exhibited by the direct, positive relationship
between the use of internal codified sources and integration centrality.
Moreover, communication with non-co-located coworkers does not have a
significant relationship with any of the other model constructs. This group
appears to rely more on the codification of their knowledge than on
interpersonal communication with non-co-located coworkers to transfer
knowledge between units. Additionally, centrality does not mediate the
relationship between internal knowledge exchange and creative performance as
hypothesized. Rather there is a direct relationship between internal knowledge
exchange and creative performance. As with the entire sample, creative
performance is also dependent upon the use of external knowledge sources as

mediated through external and internal knowledge trading.
One explanation for these findings is that individual performance for the
Design Group is much more dependent upon the ability to meet local market
demands. Creating successful front-end internet solutions is dependent upon
how end users behave when using information technology. Usage of the
internet has been found to differ across national cultures, thus this might limit
the degree to which solutions developed in one unit of the multinational may be
transferred to another multinational in the firm. Thus, what may be more
important for creative performance than accessing knowledge from other units
is the use of local external sources to gather knowledge and then the
recombination of this external knowledge with local internal knowledge
through internal knowledge trading with co-located coworkers. Integration
centrality in the firm as a whole is dependent upon the use of internal codified
sources. However, we do not find a negative relationship between the use of
external knowledge sources and integration centrality as we do with the entire
sample. Individuals in the Design Group who are more central in terms of
knowledge integration for the multinational are not those who exhibit a higher
level of creative performance. Thus, while they may be useful in terms of
knowledge transfer within the organization, this is not a necessary condition for
higher performance.
System and Software Group

As for the System and Software Group, this group is in strong contrast to the
other two groups of employees in this organization. This group shows a very
high degree of dependency on external sources of knowledge and in particular
codified knowledge sources. While for the entire sample as a whole, we find a
negative relationship between the use of external knowledge sources and
integration centrality, this relationship is positive for the System and Software
Group. In addition, the use of internal codified sources is related to a lower
degree of creative performance.
One explanation for this finding may be linked to the nature of the
system and software development field. The pace of change may be so fast
within this field that in order for individuals to keep pace with development,
they must rely to a high degree on the use of external knowledge sources.
Knowledge within the firm may quickly become out of date and the reuse of
this “old” knowledge may hamper the ability to develop and implement new
solutions and processes. In addition, system and software engineering and
design have been argued to be similar to traditional R&D (Hauptman, 1986),
which is characterized by a universal language (Allen, Tushman & Lee, 1979).
This universal language enables communication across a firm’s organizational
boundaries, facilitating the acquisition and transmission of knowledge through
a codified, textual means with individuals outside of the firm. As a result,
436 ARTICLE SIX
external knowledge used by the System and Software Group may not need to
be absorbed through the combination with internal knowledge through
knowledge exchange to the same degree as the knowledge used by the
Commercial and Support and Design Groups, thus resulting in the direct
positive relationship between the use of external sources and integration
centrality.
Additionally, we find no direct relationship between communication
with non-co-located coworkers and integration centrality. Rather this
relationship is mediated by internal knowledge exchange. Thus, for this group,
a different set of norms may be prevalent for non-face-to-face relationships
than for the other groups. Again this may be due to the universal nature of the
knowledge used as well as the ability to transfer knowledge through codified
sources by technically oriented individuals.
Implications for Theory

We find that high-performing flexible individuals are those who trade
knowledge to a high degree with individuals both within the firm as well as
outside of the firm in their networks of practice. This activity leads to their
being central within the multinational as a whole and thus their ability to
combine and recombine knowledge that is geographically dispersed across the
firm. However, high performing efficient individuals are those who merely
rely on communication with co-located coworkers. Our findings thus raise a
number of very interesting theoretical issues.
Knowledge-based View of the Firm and Knowledge Integration

These results have further enriched our view of a firm’s knowledge integration
processes and provide insight into how firms may create structures that balance
efficient and flexible knowledge integration among individuals. We find that
individuals throughout the firm perform various knowledge integration
activities through participation in networks of emergent relationships and that
an individual’s activities are related to individual performance. Relying
primarily on local knowledge sources may result in higher efficient
performance due to the ability to effectively share tacit knowledge while
creative performance is related to the ability to recombine knowledge from
different competencies as well as the ability to access non-redundant
knowledge from across and outside the organization. However, these patterns
are contingent upon the task that the individual performs.
Our findings suggest that theories of knowledge integration need to be
further developed by incorporating a dimension of cooperation. Grant’s theory
assumes that individuals are willing to share knowledge with each other
without expecting anything in return. In Hansen’s perspective, units search for
and then receive knowledge from other units through knowledge search and
transfer processes. However, our research shows that knowledge flows at the
individual level are not unidirectional as in knowledge transfer. Rather

knowledge exchange is a key element of knowledge integration – individuals
not only search for and receive knowledge from others but they are integral as
providers of knowledge to others as well. Individuals mutually engage with
each other in reciprocal knowledge sharing actions. This exchange then leads
to the combination and recombination of knowledge within the firm to enhance
flexible knowledge integration.
In addition, our research has implications for the boundaries of the firm.
The traditional view of organization design proposed that interactions with the
environment take place through formal inter-organizational relations or through
formal boundary spanning roles. However, our results indicate that individuals
are embedded not only in intra-organizational networks of practice, but they are
also embedded in networks of practice that extend across the multinational’s
legal boundaries. Individuals regardless of hierarchical level and task rely to a
high degree on their informal external contacts for advice and knowledge in
solving their everyday work tasks. This implies that the boundaries of the firm
are porous as individuals have access to advice and the latest knowledge within
their practice through participation in inter-organizational networks of practice.
In the process, individuals access external knowledge and combine it with
internal knowledge through internal knowledge exchange.
Our findings also support taking a differentiated view of firms over a
unitary one. By moving the level of analysis down to the individual, we find
that imposing one view of knowledge integration on an organization masks
possible intra-organizational heterogeneity. If we had merely looked at
performance in general or at the multinational as a whole, differences in
patterns of relationships across the groups of employees as defined by their
tasks as well as the relationship between these patterns and the two dimensions
of performance would not have been revealed. Our findings provide support
then for taking into consideration differences in the knowledge bases of firms.
Applying generic theories of knowledge integration across all firms may not be
possible, rather the appropriate one may be contingent upon the underlying
knowledge bases within the firm.
Theories of the Multinational

Looking at the multinational, our findings support Hansen’s work (1996) and
suggest that theories of the multinational structure, such as the M-form
(Chandler, 1962; Williamson, 1975), may be complemented with theories of
knowledge integration. At the local unit level, there are face-to-face
communities of practice. However, individuals within these local communities
are also members of networks of practice that span the multinational’s intra-
organizational boundaries. These individuals serve as bridges between the
local communities, exchanging and transferring knowledge between them
through these emergent relationships. In addition, individuals are also
438 ARTICLE SIX
members of inter-organizational networks of practice that tie together

individuals from a variety of external organizations. Our findings suggest that
the multinational integrates the diverse, specialized knowledge of individuals
throughout the multinational as well as accesses and integrates the knowledge
of individuals external to the firm. The multinational serves as a vehicle for the
storage of codified and tacit knowledge relating to how to integrate the
different bodies of specialized knowledge within the firm in order to ensure the
continuous efficient integration of knowledge. In addition, this storage of
knowledge relates to how the firm creates new knowledge through the
recombination of existing internal knowledge as well as through the access and
combination of external knowledge with existing internal knowledge, ensuring
the flexible integration of knowledge.
Additionally, our findings provide preliminary evidence that within the
multinational’s emergent networks, there are central individuals who are
critical to the firm’s ability to create sustainable competitive advantage. These
central individuals are key individuals in the efficient and flexible knowledge
integration processes of the firm since they influence both the organization’s
informal knowledge flows between units and the knowledge integration
processes of other individuals. These individuals may provide their own
knowledge or provide pointers to relevant experts in the firm or even outside
the firm when others from across the multinational come to them for help. As
such, they may be one of the primary determinants of the direction of the firm’s
knowledge development.
In terms of the network of practice literature, our results suggest that the
relationship between participation in various networks of practice and
individual performance is not only contingent upon the strength of the tie but
also upon the redundancy of the knowledge in the network. Our findings
suggest that the strong ties of communities of practice have a positive impact
on members’ efficient performance. Building on previous research (Teigland &
Wasko, 2002), we argue that the redundancy of the knowledge within a
community of practice in terms of functional competencies impacts members’
creative performance. Thus, the more a community of practice is
characterized by a diversity of functional competencies, the more likely that
this community of practice is able to develop more creative solutions through
the recombination of these diverse competencies. However, for communities
of practice characterized by the same functional competency, the more likely
this community of practice may turn into a competency trap or core rigidity
(Levitt & March, 1988; Leonard-Barton, 1992) unless members of this
community of practice also participate in distributed or electronic networks of
practice in which they may access knowledge that is non-redundant.
Similar to implications for the knowledge-based view of the firm, our

findings also support taking a differentiated view of networks of practice over a
unitary one. By conducting the analyses on the three task groups separately,
we find that imposing one view on a network of practice masks possible
heterogeneity. As described above, we find that network of practice centrality
is positively related to the use of external knowledge sources for the SSW
Group, but negatively related to the use of external knowledge sources for the
entire sample. Thus, applying a generic theory of networks of practice may not
be possible, rather the appropriate design may be contingent upon the
underlying knowledge base within the network of practice.
Firm Performance
Future research should focus on the relationship between firm performance and
the degree to which individuals are linked into the external environment
through participation in inter-organizational networks of practice. Previous
research suggests that one ramification of the increase in the use internet-based
communications is that it is highly likely that the amount of knowledge
exchange between organizations will increase substantially (Cronin &
Rosenbaum, 1994). The internet facilitates the rapid transmission of large data
files across corporate boundaries without any geographic, disciplinary, or
professional constraints (Wellman et al., 1996). Recent studies of knowledge
exchange are extremely limited, but in a study of one Usenet technical
community, it was found that 42% of all messages sent included programming
code (Wasko & Faraj, 1999). In our research, we find that individuals using
external knowledge sources also participate in external knowledge exchange –
sending out the firm’s internal knowledge in exchange for external knowledge.
Additionally, our unexpected finding of the positive, direct relationship
between the use of internal codified sources and external exchange provides an
indication that individuals are accessing codified firm knowledge to exchange
with external sources.
As mentioned above, previous research has revealed that this knowledge
may even include proprietary firm knowledge (von Hippel, 1987; Schrader,
1991). Thus, the activity of this inter-organizational knowledge exchange by
individuals at all levels and positions of the firm draws into question the degree
to which a company’s proprietary knowledge is leaking across its legal
boundaries (Mansfield, 1985). The decision to exchange knowledge or not
with external parties is placed in the hands of an individual working for the
firm. As such, economic and management researchers generally argue that this
informal “leakage” may be a disadvantage for the firm due to the potential
dilution of a firm’s competitive advantage (Schrader, 1991). Individuals may
trade valuable knowledge for purely personal objectives or may make a mistake
and misjudge the value of this knowledge (von Hippel & Schrader, 1996).
Research on the relationship between know-how trading and firm performance
440 ARTICLE SIX
is scant, primarily because it is very difficult for firms to manage and evaluate
the benefits since it occurs “off the books” (Carter, 1989). However, there is
some initial evidence of a positive relationship (Schrader, 1991).
Our results from this research also suggest a positive relationship since
we find that the use and exchange of external knowledge has an indirect
relationship to creative performance through their influence on internal
knowledge exchange and integration centrality. A potential explanation for our
results may be that knowledge coming from outside the firm may be so novel
that it cannot be applied to any immediate solution. Rather its dissemination to
others and subsequent recombination with the firm’s knowledge is necessary to
adapt this knowledge to the firm’s specific use. Thus, the ability to develop
and implement innovative solutions and improve performance may involve
combining existing internal knowledge with novel external knowledge. This
finding supports theories of absorptive capacity, which suggests that the firm’s
ability to assimilate new, external information is largely a function of the firm’s
ability to internally process that information (Cohen & Levinthal, 1990). As a
result, participation by individuals in inter-organizational networks of practice
may not necessarily negatively impact a firm’s competitive advantage. While
proprietary practice knowledge may leak out in these inter-organizational
networks, the ability of another organization to use this knowledge to its
advantage then becomes dependent upon its ability to absorb and adapt it to its
specific use through internal knowledge exchange.
Implications for Practice
Knowledge Management
One of the primary areas of implications of the results from our research is for
the field of knowledge management. Our findings indicate that organizations
concerned with knowledge management may need to rethink their knowledge
management strategies. First, what do we make of the result that different
patterns of knowledge integration are associated with different performance
outcomes? The first is the balance between efficient and flexible integrative
performance. With the rapid development of the ease of use of the internet
(e.g., smart agents, more specialized discussion forums) and the increasing
ability of individuals to use the internet and communicate with others in their
external networks, this media will become a much more helpful knowledge
source. Yet, as we found, it is communication with co-located coworkers that
leads to superior efficient performance. The question then becomes how to
balance the use of the new media to ensure a productive ratio of flexibility to
efficient performance. As we find here, knowledge integration patterns differ
depending upon which type of performance is the objective. In some
organizations, a focus on efficient performance through systems that promote
local communication may be the objective while in others a focus on flexible
performance through systems that promote the development of intra-

organizational networks of practice may be the objective. Thus, knowledge
management systems should be aimed at facilitating a balance between
efficient and creative performance that matches a company’s competitive
strategy.
Second, what is to be made of the high use of external knowledge
sources? To date, knowledge management systems have focused on leveraging
knowledge within the organization. In some instances, these systems also may
include other organizations (extranets); however, these other organizations
often fall within the realm of the organization’s formal task environment. As
we see here, the use of informal external sources is indirectly related to creative
performance. A challenge then for management is whether knowledge
management systems should be developed that facilitate the use of external
sources. In previous research at Icon (Teigland, 2000), interviews with
individuals in the System and Software Group revealed that these individuals
prefer to go outside the organization to external electronic communities to
asking others within Icon for help. Not only can more answers be found to
their question but the speed with which answers are given is much faster than
using internal knowledge sources.
Third, this research shows that implementing generic knowledge
management strategies across an organization may not prove successful. The
three groups of Commercial and Support, Design, and System and Software
differ significantly in their activities related to knowledge integration and their
relationship to individual performance. These differences imply that
knowledge management systems tailored to each group of employees based on
its practice knowledge may be more successful. For example, for groups such
as the Design Group at Icon, resources may be better spent on developing
internal document repositories; however, for groups such as the System and
Software Group, resources may be better spent on leveraging the use of
external knowledge sources as well as ways to improve knowledge flows
within the internal network.
Organizational Identity and Commitment

Just as individuals have a certain degree of commitment to their organizations,
they also have a degree of commitment to their profession or occupation as
several researchers have noted (Arrow, 1974; Saxenian, 1996; Brown &
Duguid, 2001). In some professions, the degree of commitment to the
profession can be so strong that the norms of the profession even transcend the
norms of the organizations that employ the individuals. Members of
professions can be separated by great distances and still see themselves as part
of the same professional group. Academics have long been examples of
individuals with a strong degree of commitment to their profession (Pickering
& King, 1995).
442 ARTICLE SIX
The implication from our results is that individuals may then hold
multiple identities, as Kogut & Zander (1996) argue. On the one hand,
individuals belong to a community within the firm, while on the other, they
may belong to a professional community outside the firm, as evidenced through
the high use of external knowledge sources. Thus, an individual may be faced
with competing allegiances and conflicting objectives. As a result, a challenge
for management is to ensure that employees have the right balance between
participation in the firm and participation in external networks of practice. If
individuals have a higher degree of loyalty to their network outside the
organization, this may be at the expense of the company, leading to two
ramifications for company performance: 1) an individual spending too much
time participating in external networks during working hours and 2) the giving
away of proprietary know-how or information.
First, individuals may be too involved with their external community
and as a result, spend too much time “working for” their community. While
not found here, previous research provides evidence that individuals who spend
a lot of time working with others in electronic networks of practice are more
likely to have a poor level of efficient performance (Teigland, 2000).
Interviews with some of these individuals revealed that they often were so busy
helping others outside their organization or striving to create elegant or
“bleeding edge” solutions that they were unable to focus on finishing their own
tasks according to management’s objectives. It appeared that these individuals
had considerable “power” over management. This power resulted from
management’s inability to understand in detail what their employees were
doing since they were unable to keep up with the rapid pace of technological
development. Thus, management often did not know whether employees were
working on necessary value-adding activities or were spending time trying to
impress their peers. One manager summarized this situation with reference to
the software programmers, “Programmers take us (management) hostage. We
never know whether they’re working on extra bells and whistles to impress
their buddies or whether it’s really a value-adding activity for the customer.”
For some individuals it may be worth more for them to develop the “latest and
coolest” solution than to complete their work on time and to their supervisor’s
requests.
As discussed under the above section, Implications for Theory, the
second challenge is more of a threat to a firm’s competitive advantage – the
trading of proprietary know-how or information. The development of the
internet has greatly facilitated the degree by which proprietary information can
be traded between companies without management’s knowledge. In many
cases, management is completely unaware that its employees are participating
in this behavior. The decision whether to transmit proprietary knowledge is
placed within the hands of the individual. If an individual is trying to enhance
his or her identity in the external community, then he or she may transmit this
proprietary information without receiving anything of value for the company in

return. However, as described above, the results of this research seem to
provide contrary evidence, that external knowledge exchange is beneficial for
the firm, although indirectly through the recombination with existing
knowledge. Thus, while an individual may trade away “proprietary”
knowledge, the ability of a rival firm to turn this into an innovation lies in its
ability to integrate it with its own existing knowledge bases in the firm.
Third, while some organizations have attempted to stop the cross-
boundary flow of knowledge within these groups, these attempts may be
counterproductive (Brown & Duguid, 2001). Attempting to block participation
in inter-organizational network knowledge flows may only lead to increased
loyalty to the external network and decreased loyalty to the organization and
potentially a negative effect on individual performance. And as noted above,
these flows are two-way, generally characterized by an equal exchange of
knowledge. Thus, of importance to the knowledge integration view, cutting off
flows to outside of the firm will more than likely risk cutting off flows into the
firm (Saxenian, 1996; Brown & Duguid, 2001).
Finally, one implication of this research is that when management hires
a person, management is also “hiring” the employee’s external network as well.
Thus, management must consider the potential employee’s external network
and how active the individual is in this network. If the person is very active in
this external network, then the individual’s time may be spent on external
activities. As indicated here, individuals who participate to a high degree only
in external knowledge exchange and not in internal knowledge exchange, are
less central in the firm. Another aspect to consider is which individuals are
included in the potential employee’s network and what value these individuals
may provide to the organization through external knowledge exchange.
Central Individuals
This research has also provided evidence that management should pay
considerable attention to understanding the emergent networks of its
organization. Through their activities related to knowledge integration,
individuals achieve different levels of efficient and creative performance.
Management should focus on developing an understanding of which
individuals are the central individuals in the firm’s emergent networks for
several reasons. First, as mentioned above, central individuals are influential in
developing the common language and norms of the local community and as
such determine to what degree the languages and norms vary between the local
communities of practice spread across the multinational. Second, central
individuals are key players in the future of the multinational. They are one of
the primary determinants in the direction of the firm’s knowledge development
since these individuals make critical decisions regarding which external
knowledge is combined with which internal knowledge. Third, while most
444 ARTICLE SIX
individuals throughout the firm participate in some kind of external knowledge

exchange, central individuals have a greater possibility of trading away the
firm’s more valuable knowledge since they have greater access to the firm’s
more valuable resources and critical knowledge. Decisions to trade and the
ability to obtain valuable external knowledge are based on their own judgment.
In summary, since knowledge is argued to be the most valuable resource of the
firm and sustainable competitive advantage stems from an organization’s
capability to integrate the knowledge of its members, central individuals in a
firm’s emergent networks are critical to the firm’s knowledge integration
processes and the ability to create sustainable competitive advantage.
Limitations and Areas for Future Research

Despite the strengths of this study involving a large, all-inclusive multinational
organization, we should also note the limitations of the study. First, this study
only examines individuals within one company, thus limiting the
generalizability of our findings. Further research should examine activities
related to knowledge integration and their relationship to individual
performance across multiple organizations. Second, data were collected at one
point in time. Another limitation is our use of self-reported survey measures
for performance only. Thus, future research should include objective data
sources in addition to survey data as well the collection of data over time to
further establish the relationship to performance.
Additionally, in this research, we focus on the full scale of knowledge
and do not incorporate the whole scale from tacit to explicit knowledge. We
also do not look at the strength of the relationships between individuals in the
advice networks. However, previous research suggests that at the unit level the
relationship between performance and the position in the firm is dependent
upon the type of knowledge being transferred as well as the strength of the tie
(Hansen, 1996). Thus, future research should incorporate the different
dimensions of knowledge as well as tie strength.
In terms of knowledge integration, Grant (1996a) proposed three
dimensions of effective knowledge integration. We investigated two of these:
efficient and flexible knowledge integration. However, due to the already
complex nature of this study, we did not investigate the third dimension, the
scope of integration. Clearly, it would be of interest to look at the effect of the
scope of knowledge integration.
Finally, this research has been conducted solely at the individual level.
However, as discussed in the Implications section, there are several
implications of our research for both the unit and the firm level. Thus, an
interesting area for further research would be to understand the relationship
between the patterns of individual level knowledge integration and unit level
knowledge integration and the effect on unit or firm performance.
Conclusion
In conclusion, this research set out to investigate the activities related to
knowledge integration at the individual level within a multinational firm. We
find indications that these activities differ depending upon whether efficient or
creative performance is the desired outcome and that individual performance is
also dependent upon the task performed. Our results provide support for the
knowledge-based view of the firm as well as indicate how these theories may
be further developed. In terms of implications, one of the most interesting
results from this research is the extent to which a firm’s external boundaries are
becoming porous. Individuals rely on the use of codified external knowledge
sources as well as their external informal networks to solve their everyday
work-related problems and in the process exchange firm knowledge to gain
access to external knowledge. This finding clearly implies that researchers and
practitioners alike need to further investigate these inter-organizational
knowledge flows and the impact that these have on competitive advantage and
firm performance.
REFERENCES
Ahuja, M.K., Galletta, D.F., Carley, K.M. 2003. Individual centrality and
performance in virtual R&D groups: An empirical study. Management
Science, 49,1: 21-38.
Aiken, M. & Hage, J. 1968. Organizational interdependence and intra-
organizational structure. American Sociological Review, 33,6: 912-930.
Aldrich, H. & Herker, D. 1977. Boundary spanning roles and organization
structure. Academy of Management Review, April: 217-230.
Allen, T.J. 1977. Managing the Flow of Technology: Technology Transfer and
the Dissemination of Technological Information Within the R&D
Organization. Cambridge, MA: MIT Press.
Allen, T.J. & Cohen, S.I. 1969. Information flows in research and development
laboratories. Administrative Science Quarterly
Allen, T.J., Tushman, M.L. & Lee, D.M.S. 1979. Technology transfer as a
function of position in the spectrum from research through development to
technical services. Academy of Management Journal, 22,4: 694-708.
Arrow, K. 1974. The Limits of Organization. New York: Norton Co.
Ashford, S.J., Rothbard, N.P., Piderit, S.K., & Dutton, J.E. 1998. Out on a
limb: The role of context and impression management in selling gender-
equity issues. Administrative Science Quarterly, 43: 23-57.
Bagozzi, R.P. & Edwards, J.R. 1998. A general approach to representing
multifaceted personality constructs: Application to state self-esteem.
Structural Equation Modeling, 1: 35-67.
Baldwin, N.S. & Rice, R.E. 1997. Information-seeking behavior of securities
analysts: Individual and institutional influences, information sources and
channels, and outcomes. Journal of the American Society for Information
Science, 58, 8: 674-693.
446 ARTICLE SIX
Bartlett, C.A. & Ghoshal, S. 1989. Managing Across Borders: The

Ben-David, J. 1969. National and international scientific communities. In D.J.
de Solla Price & V. Voise (eds.) National Scientific Communities: A
Sociological Study of Developing and Developed Countries. Paris: United
Nations Educational, Scientific and Cultural Organization.
Ben-David, J. 1971. The Scientist’s Role in Society: A Comparative Study.
Englewood Cliffs, NJ: Prentice-Hall.
Bentler, P.M. 1980. Multivariate analysis with latent variables: Causal
modeling. Annual Review of Psychology, 31: 419-456.
Bentler, P.M. & Bonett, D.G. 1980. Significance tests and goodness of fit in
analysis of covariance structures. Psychological Bulletin, 88: 588-606.
Bernard, R.H, Killworth, P., Kronenfeld, D. & Sailer, L. 1984. The problem of
informant accuracy: The validity of retrospective data. In B.J. Siegel (ed.).
& A.R. Beals & S.A. Tyler Assoc. (eds.), Annual Review of Anthropology,
13: 495-517. Palo Alto, Cal: Annual Reviews.
Blau, P. M. (1964). Exchange and Power in Social Life. New York: Wiley.
Bommer, W.H., Johnson, J.L, Rich, G.A., Podsakoff, P.M., & MacKenzie, S.B.
1995. On the interchangeability of objective and subjective measures of
employee performance: A meta-analysis. Personnel Psychology, 48,3: 587.
Borgatti, S.P., Everett, M.G., & Freeman, L.C. 1992. UCINET IV Version 1.0
Reference Manual. Columbia, SC: Analytic Technologies.
Borman, W.C. 1978. Exploring upper limits of reliability and validity in job
performance ratings. Journal of Applied Psychology, 63,2: 135-144.
Bouty, I. 2000, Interpersonal and interaction influences on informal resource
Academy of Management Journal, 43, 1: 50-65.
Brown, J.S. & Duguid, P. 2000. The Social Life of Information. Boston:
Burkhardt, M. E., & Brass, D. J. 1990. Changing patterns of patterns of change:
The effects of a change in technology on social network structure and power.
Administrative Science Quarterly, 35, 104-127.
Chandler, A. 1962. Strategy and Structure: Chapters in the History of the
Industrial Enterprise. Cambridge, MA: MIT Press.
Carter, A.P. 1989. Knowhow trading as economic exchange. Research Policy,
18: 155-163.
Cohen, W. M. & Levinthal, D. A. 1990. Absorptive capacity: A new
perspective on learning and innovation. Administrative Science Quarterly,
35: 128-152.
Science, 7: 119-135.
Cronin, B. & Rosenbaum, H. 1994. Netelligence: Informal know-how trading

on the internet. International Journal of Information Management, 14: 144-
145.
Cudeck, R. & Browne, M.W. 1983. Cross-validation of covariance structures.
Multivariate Behavioral Research, 18: 147-167.
Demsetz, H. 1988. The theory of the firm revisited. Journal of Law,
Economics, and Organization, 4: 141-161.
Doz, Y. & Hamel, G. 1997. Alliance Advantage. Cambridge, MA: Harvard
Faraj, S. & Wasko, M. M. 1998. The Web of Knowledge: An Investigation of
Self-Organizing Communities of Practice on the Net. Paper presented at the
INFORMS, Seattle, Washington.
Feldman, M. 1987, Electronic mail and weak ties in organizations. Office:
Technology and People, 3, 3: 83-101.
Freeman, L.C., Romney, A.K. & Freeman, S.C. 1987. Cognitive structure and
informant accuracy. American Anthropologist, 89: 310-325.
Gerstberger, P. & Allen, T.J. 1968. Criteria used by research and development
Psychology, 52: 272-279.
Sociology, 91: 481-510.
7,4: 375-387.
Grant, R. M. 1996b. Toward a knowledge-based theory of the firm. Strategic
Management Journal, 17 (Special Issue): 109-122.
Hansen, M.T. 1996. Knowledge Integration in Organizations. Ph.D.
Dissertation, Graduate School of Business, Stanford University.
Hansen, M.T. 1999. The search-transfer problem: The role of weak ties in
sharing knowledge across organization subunits. Administrative Science
Quarterly, 44: 82-111.
Harris, M.M & Schaubroeck, J. 1988. A meta-analysis of self-supervisor, self-
peer, and peer-supervisor ratings. Personnel Psychology, 41: 43-62.
Hauptman, O. 1986. The influence of task type on the relation between
communication and performance: The case of software development. R &
D Management, 16,2: 127-139.
Hedlund, G. 1994. A model of knowledge management and the N-form
corporation. Strategic Management Journal, 15.
Hedlund, G. & Nonaka, I. 1993. Models of knowledge management in the
West and Japan. In: Lorange, P., Chakravarthy, B., Roos, J. & Van de Ven,
A., (eds.). Implementing Strategic Process, 117-145. Blackwell Business.
Hinds, P. & Kiesler, S. 1995. Communication across boundaries: Work,
Organization Science, 6: 373-393.
448 ARTICLE SIX
Kettinger, W.J. & Grover, V. 1997. The use of computer-mediated

communication in an interorganizational context. Decision Sciences, 28, 3:
513-555.
Kogut, B. & Zander, U. 1996. What firms do? Coordination, identity and
learning. Organization Science, 7,5: 502-518.
Kollock, P & Smith, M. 1996. Managing the virtual commons: Cooperation
and conflict in computer communities. In S. Herring (ed.), Computer-
mediated Communication: Linguistic, Social, and Cross-Cultural
Perspectives. Amsterdam: John Benjamins.
Lakhani, K. & von Hippel, E. 2000. How open source software works: Free
user-to-user assistance. Paper presented at The 3rd Intangibles Conference.
Business, NYU.
Leifer, R. & Huber, G.P. 1977. Relations among perceived environmental
uncertainty, organizational structure, and boundary-spanning behavior.
Levitt, B. & March, J. G. 1988. Organizational learning. Annual Review of
Sociology, 14: 319-340.
Liefeld, J. P. 1988. Response effects in computer-administered questioning.
Journal of Marketing Research, 25, 4: 405-409.
Macdonald, S. 1995. Learning to change: An information perspective on
learning in the organization. Organization Science, 6,5: 557-568.
Macdonald, S. & Williams, C. 1993. Beyond the boundary: An information
perspective on the role of the gatekeeper in the organization. Journal of
Production Innovation Management, 10: 417-427.
Martinez, J.I. & Jarillo, J.C. 1989. The evolution of research on coordination
mechanisms in multinational corporations. Journal of International
Business Studies, Fall: 489-514.
Marsden, P. 1990. Network data and measurement. In W.R. Scott& J. Blake
(eds.). Annual Review of Sociology, 16: 435-463.
Mehra, A., Kilduff, M., & Brass, D.J. 2001. The social networks of high and
low self-monitors: Implications for workplace performance. Administrative
Nahapiet, J. & Ghoshal, S. 1998. Social capital, intellectual capital, and the
Newsted, P.R. 1985. Paper versus online presentations of subjective
questionnaires. International Journal of Man-Machine Studies, 23, 1: 231-
247.
Nonaka, I. & Takeuchi, H. 1995. The Knowledge Creating Company, New

York: Oxford University Press.
O'Reilly, C. 1982. Variations in decision makers' use of information sources:
The impact of quality and accessibility of information. Academy of
Management Journal, 25, 4: 756-771.
Orlikowski, W. J. (1996). Learning from Notes: Organizational issues in
groupware implementation. In R. Kling (Ed.), Computerization and
Controversy: Value Conflicts and Social Choices (pp. 173-189). New York:
Academic Press.
Cornell University.
Pickering, J. M. & King, J. L. 1995. Hardwiring weak ties: Interorganizational
Polanyi, M. 1962. Personal Knowledge: Towards a Post-Critical Philosophy.
Chicago: University of Chicago Press.
Reber, A.S. 1993. Implicit Learning and Tacit Knowledge, New York: Oxford
University Press.
Saxenian, A. 1996. Regional Advantage: Culture, competition in Silicon
Valley and Route 128, Second edn. Cambridge, MA: Harvard University
Press.
Schenkel, A. 2002. Communities of Practice or Communities of Discipline:
Managing Deviations at the Öresund Bridge. Published Ph.D. dissertation,
Stockholm School of Economics.
Schenkel, A, Teigland, R. & Borgatti, S.P. 2001. Theorizing structural
properties of communities of practice: A social network approach. Academy
of Management Conference.
Seibert, S.E., Kraimer, M.L., & Liden, R.C. 2001. A social capital theory of
career success. Academy of Management Journal, 44, 2: 219-237.
Snyder, W. 1996. Organizational Learning and Performance: An Exploration
of Linkages Between Organizational Learning, Knowledge, and
Performance. University of Southern California. Ph.D. Dissertation.
Sparrowe, R.T., Liden, R.C., Wayne, S.J., & Kraimer, M.L. 2001. Social
networks and the performance of individuals and groups. Academy of
technology. In B. K. J. Keller (ed.), Public Access to the Internet: 62-81.
Cambridge, MA: MIT Press.
Sproull, L. & Kiesler, S. 1986. Reducing social context cues: Electronic mail
in organizational communication. Management Science, 32: 1492-1512.
Steiger, J.H. 1990. Structural model evaluation and modification: An interval
estimation approach. Multivariate Behavioral Research, 25,2: 173-180.
450 ARTICLE SIX
Szulanski, G. 1996. Exploring internal stickiness: Impediments to the transfer

of best practice within the firm. Strategic Management Journal, 17: 27-43.
Taylor, R.L. 1975. The technological gatekeeper. R & D Management, 5: 239-
242.
Teigland, R. 2000. Communities of practice at an internet firm: Netovation vs.
on-time performance. In E. L. Lesser, M.A. Fontaine, & J.A. Slusher (eds.),
Knowledge and Communities: 151-178. Boston: Butterworth-Heinemann.
Teigland, R. & Wasko, M.M. 2002. The Provision of Online Public Goods:
Examining Social Structure in a Network of Practice. In The Proceedings
of the 23rd International Conference on Information Systems,
Barcelona, Spain, ICIS.
Teigland, R. & Wasko, M.M. 2003a. Extending richness with reach:
Participation and knowledge exchange in electronic networks of practice. In
P.M. Hildreth (ed.). Knowledge Networks: Innovation Through Communities
of Practice. London: Idea Group Inc. Forthcoming.
Teigland, R. & Wasko, M.M. 2003b. Integrating knowledge through
information trading: Examining the impact of boundary spanning
communication on individual performance. Decision Sciences, forthcoming
(Special Issue on Knowledge Management).
Thibaut, J. W., & Kelley, H. H. (1959). The Social Psychology of Groups. New
York: John Wiley.
Tsai, W. & Ghoshal, S. 1998. Social capital and value creation: The role of
intrafirm networks. Academy of Management Journal, 41,4: 464-476.
Tsoukas, H. 1996. The firm as a distributed knowledge system: A
constructionist approach. Strategic Management Journal17: 11-25.
Turban, D.B. & Dougherty, T.W. 1994. Role of protégé personality in receipt
of mentoring and career success. Academy of Management Journal, 37, 3:
688-702.
Science, 26,11:1071-1085.
Tushman, M.L. & Scanlan, T.J. 1981. Characteristics and external orientations
of boundary spanning individuals. Academy of Management Journal,
24,1:83-98.
Van de Ven, A.H. 1976, Delbecq, A.L., Koenig, R. A framework for
organizational assessment. American Sociological Review, 41: 322-338.
Van Maanen, J. & Barley, S. R. 1984. Occupational communities: Culture and
control in organizations. In B. M. Staw & L. L. Cummings (eds.). Research
in Organizational Behavior, Vol. 6. Greenwich, CO: JAI Press.
Von Hippel, E. 1987. Cooperation between rivals: Informal know-how trading.
Von Hippel, E. & Schrader, S. 1996. Managed informal information trading:
Walker, G. 1985. Network position and cognition in a computer software firm.
Administrative Science Quarterly, 30,1: 103-131.
Wasko, M. & Faraj, S. 1999. Webs of knowledge: An examination of

practice. Paper presented at the Academy of Management Conference,
Chicago, Ill.
Wasserman, S. & Faust, K. 1994. Social Network Analysis: Methods and
Applications. Cambridge: Cambridge University Press.
Weekley, J.A. & Gier, J.A. 1989. Ceilings in the reliability and validity of
performance ratings: The case of expert raters. Academy of Management
Journal, 32, 1: 213-222.
Welbourne, T.M., Johnson, D.E., Erez, A. 1998. The role-based performance
scale: Validity analysis of a theory-based measure. Academy of
Wellman, B., Salaff, J., Dimitrova, D., Garton, L., Gulia, M., &
Haythornthwaite, C. 1996. Computer networks as social networks:
Collaborative work, telework, and virtual community. Annual Review of
Sociology, 22: 213-238.
Williamson, O.E. 1975. Markets and Hierarchies: Analysis and Antitrust
Implications: A Study in the Economics of Internal Organization. New
York: The Free Press.
Yanow, D. 2000. Seeing organization learning: A cultural view. Organization,
7,2: 247-268.
Zander, U. & Kogut, B. 1995. Knowledge and the speed of the transfer and
imitation of organizational capabilities: An empirical test. Organization
Science, 6,1: 76-92.
Article Seven
Knowledge Dissemination in Global R&D

Operations: An Empirical Study of Multinationals in
the High-Technology Electronics Industry
We use the term, community of practice, to represent networks of practice both

within the firm and outside the firm.
454 ARTICLE SEVEN
Published in Management International Review, Special Issue on International

Management of Technology, 2000, 1: 49-78
Knowledge Dissemination in Global R&D

Operations: An Empirical Study of Multinationals in
the High-Technology Electronics Industry
Carl Fey
Stockholm, Sweden
email: carl.fey@hhs.se
Robin Teigland
Stockholm, Sweden
Julian Birkinshaw
London Business School, London, England
email: jbirkinshaw@london.edu
ABSTRACT
This study is an empirical examination of 1) the mechanisms that three
multinationals used to disseminate knowledge within their globally dispersed
R&D operations, 2) the impediments that these multinationals experienced
when implementing these mechanisms, and 3) the means to overcome these
impediments. Management focused considerably on implementing
mechanisms that facilitated the flow of knowledge to top management for the
coordination of globally dispersed R&D activities. However, little attention
was placed on mechanisms to facilitate the flow of knowledge for use by
engineers in problem-solving activities. Three types of impediments to
knowledge flow were observed: 1) opportunity cost of time, 2) knowledge is
power, and 3) not-invented-here. Knowledge flow in an MNC appears to be
facilitated by establishing a one-company culture through 1) incorporating
teamwork as a company value, 2) evaluating individual knowledge contribution
and assimilation in performance appraisals, 3) implementing a goal that
promotes overall company improvement, and 4) facilitating extensive
personnel rotation.
Keywords: multinational, community of practice, performance, R&D,

knowledge
KNOWLEDGE DISSEMINATION IN GLOBAL R&D OPERATIONS 455
KNOWLEDGE DISSEMINATION IN GLOBAL R&D

OPERATIONS
Multinational corporations (MNCs) have largely built their success on their
ability to create competitive advantage through technological innovations.
Traditionally, MNCs operated with a centralized R&D structure that facilitated
the creation of new technology in the home country followed by subsequent
technology dissemination to overseas operations. Over the past couple of
decades, however, the competitive environment has changed, and as a result,
MNCs have had to re-think their approach to innovation and R&D
management. Now, it is argued that to achieve competitive advantage MNCs
must continuously create, transfer, and exploit knowledge that is increasingly
dispersed throughout their global operations (Bartlett & Ghoshal, 1989;
Hedlund & Nonaka, 1993; Doz et al., 1997).
The traditional centralized R&D structure that was seen in the post-war
years is gradually being eclipsed by the emergence of global R&D networks
(Ronstadt, 1977; Terpstra, 1977; Lall, 1979; Håkanson, 1990). Four main
reasons seem to be underlying this structural change:
1) Number of knowledge sources increasing. Companies are establishing

R&D centers in multiple locations to tap into and absorb the new
knowledge and research produced by universities and competitors across
the globe (de Meyer, 1989; Kuemmerle, 1997).
2) Localization of competition. As global competition increases, the ability
to win market share in many industries lies also in the ability to adapt to
local needs. Centralized R&D may not be able to provide the locally
needed solutions in the most efficient manner (Mansfield et al., 1979; de
Meyer, 1989).
3) Shrinking product lifecycles. As the time from development to market
continues to shrink, companies are establishing units in the marketplace
that enable the company to rapidly move the product through the
commercialization phase (de Meyer, 1989).
4) Diversity of skills and technologies required. The diversity of skills and
technologies necessary to produce a certain good or service is rising.
These are increasingly found across the globe at multiple sites. In
addition, these multiple sites encourage the development of more ideas
due to the varied international backgrounds in global networks (Barth et
al., 1995).
Managing R&D Networks

Globally dispersed networks of R&D units create significant managerial
challenges to MNCs. The basic challenge is one of maintaining the
456 ARTICLE SEVEN
responsiveness of individual units to the opportunities and demands of their

local environment while at the same time capturing the latent benefits that a
large, global network can confer (Bartlett & Ghoshal, 1989). Knowledge, it is
argued, must be created at a lightening pace while it is simultaneously
transferred and applied throughout the global operations. Coordination across
units prevents the duplication of effort while at the same time ensuring the
fastest time to market with a product that customers want. However, this quick
pace leads to the need for unit autonomy since decisions must be made rapidly
over distributed locations (Håkanson & Zander, 1986; Håkanson, 1990).
The task of efficiently making use of R&D knowledge becomes more
difficult as many MNCs continue to expand their global R&D operations, and
thereby increase the number of geographically dispersed locations, employees,
functions, and external partners. Both the complexity of the network and the
differences in language and culture lead to significant challenges.
How are MNCs responding to the increased demands on their global
R&D operations? The standard argument in the literature is that firms are
building “integrated networks” (Håkanson and Zander, 1986), in which task
specialization occurs at a group, project, or unit level, but it is overlain with a
variety of integrating mechanisms that ensure rapid and effective flow of
knowledge across units. However, while such a model is attractive in theory, it
is also recognized that effective knowledge flows are hard to achieve (Behrman
& Fischer, 1980; de Meyer, 1991, Granstrand, Håkanson & Sjölander, 1992;
Nobel & Birkinshaw, 1998).
The aim of this paper, then, is to focus on an applied question: How are
MNCs actually managing the knowledge flows in their global R&D networks?
The question is descriptive, but it builds on our a priori expectation that most
firms would like to see an increase in both the volume and quality of
knowledge flows between and within R&D units. As such, our study considers
both the facilitators of and the obstacles to knowledge flow.
It is important to be clear on the positioning of this research vis-à-vis
other studies in the same broad area. In the R&D management literature,
patterns of communication have been well studied, both within a single site
(Allen, 1977), and between R&D units (de Meyer, 1991; Nobel & Birkinshaw,
1998), but patterns of knowledge flow have not. In the rapidly growing
knowledge management literature (see below), knowledge flows have of course
been studied. However, from our reading of the literature, papers have tended
to focus on conceptual models (Hedlund, 1994; Nonaka & Takeuchi, 1995;
Grant, 1997) or discrete transfer events (e.g., Szulanski, 1996; Zander, 1991).
Relatively little attention has been given to the full scope of knowledge flows
that are to be found within a single organizational setting. This paper, then,
hopes to contribute to both literatures by providing detailed, multi-level
evidence of knowledge flows in the R&D organizations of three MNCs.
KNOWLEDGE MANAGEMENT
The focus on knowledge flows in this paper makes it necessary to briefly
consider what we mean by knowledge. We recognize first of all that there is a
“software” and a “hardware” side to knowledge73. The hardware side or the
articulate form of knowledge is that which is represented explicitly in physical
or material objects such as a patent. It is the know-what or information (Kogut
& Zander, 1992). Tacit knowledge (Polanyi, 1969; Hedlund & Nonaka, 1993),
the software side, is intuitive, non-verbalized, and not yet articulated. It is the
"know-how" or the practical skills or expertise that allows a researcher to work
smoothly and efficiently (von Hippel, 1988; Kogut & Zander, 1992). Hedlund
and Nonaka (1993) claim that the creation of knowledge occurs then when tacit
knowledge is articulated through the codification of experience and information
into articulate form.
In terms of knowledge flow or dissemination, some local knowledge is
more articulated and thus more easily transferable across borders. Mechanisms
that lend themselves to the transfer of articulate knowledge are broad channel
communication or automated information distribution methods, databases, or
groupware. However, research indicates that more than half of the knowledge
in organizations is tacit and an even greater portion of the valuable knowledge
is tacit (Snyder, 1996, 1997). This tacit knowledge is locally specific and
harder to get access to from a distance (Westney, 1993). Thus, it is more costly
to be transferred to other parts of the world (von Hippel, 1988; Asakawa,
1995).
How then is it possible to transfer or disseminate tacit knowledge? It
has been argued that the most effective means is actually not to codify it.
Rather than attempt to codify knowledge through IT systems, studies suggest
that a more effective means is to involve people in face-to-face interactive
activities such as storytelling, dialogue, and peer coaching which facilitate the
learning of beliefs (Brown & Duguid, 1991; Snyder, 1996, 1997). Such
informal interactive activities are the hallmarks of “communities of practice”
which are defined as a set of people who are bound by informal relationships
and share a common practice (Snyder, 1996; Wenger, 1998). The people in
these communities are often not bound to typical geographic, SBU, or
functional boundaries in organizations. Instead they are linked through
informal practice and personal-based networks (Wenger, 1998; Snyder, 1997).
While there are great benefits to the firm in effectively transferring knowledge
internally, there are also risks. Making knowledge easy to transfer is a double-
edged sword, because the characteristics that facilitate knowledge transfer
inside the firm –articulability, observability, system independence – are also
likely to make it relatively easy for competitors to imitate (Kogut & Zander,
73
This division resembles the research conducted by Blakeslee (1985) on brain structures in which he
distinguished between those structures used for the memory of the declarative knowledge of facts and
those structures used for the memory of procedural knowledge that underlies skills.
458 ARTICLE SEVEN
1992). This is a critical issue, especially as one gets into the link between firm
knowledge and sustainable competitive advantage, but it is not one that we
address in detail in this paper. However, Kogut and Zander’s (1992) argument
that firms outperform markets as vehicles for tacit knowledge transfer is one
important guard against competitive imitation, and it is also consistent with our
findings regarding the importance of a one-company culture.
CONCEPTUAL FRAMEWORK
How MNCs manage their R&D knowledge flows is a broad area of inquiry, so
our purpose in this section is to focus the question somewhat and put forward
the organizing framework that we used to structure our investigation. Two
specific restrictions on our investigation should be made at the outset. First,
while knowledge flows can be concerned with both creation and dissemination
(voluntary and involuntary), our focus is explicitly on the dissemination
process. Second, we do not consider the flows of knowledge beyond the
boundaries of the firm. As noted above, this is an important issue but it is
beyond the scope of the current paper.
There are numerous ways to classify dissemination mechanisms but for
the purpose of this research, a simple matrix was developed (see table 1). The
vertical axis refers to the basic type of mechanism used to disseminate
knowledge (IT application or organizational measures). Organizational
measures are based upon face-to-face interactions such as personnel rotation or
cross-functional teams. The horizontal axis of the matrix is based upon the task
for which the knowledge that is being disseminated will be used, e.g., problem-
solving vs. coordinating.
Table 1 Knowledge Dissemination Classification Matrix

Problem-solving Problem-solving Coordinating
Individual Group Organizational
IT Applications
Organizational Measures
This matrix allows us to focus our broad research question. The first
objective of the study falls straight out of the framework presented in figure 1,
namely identifying the various mechanisms that MNCs are using to enhance
knowledge flows, as they pertain to individual problem-solving, group
problem-solving and coordinating, and organizational coordination. More
formally:
RESEARCH QUESTION 1: What mechanisms are MNCs using to

transfer knowledge in their global R&D operations?
The second objective of the study is to give explicit attention to the

obstacles to knowledge flows. Achieving the desired level of knowledge flows
is not easy. The successful implementation of any knowledge management
technique depends on the willingness and enthusiasm of its individual
employees to adopt it. Pasmore & Gurley (1991) refer to this as a lack of
cooperation claiming that people withhold knowledge if they are in a
competitive relationship, feel wronged, could lose political power, or if they
feel that the knowledge is not of use to anyone. Davenport & Prusak (1998)
discuss the lack of trust and time and the Not-Invented-Here syndrome as
frictions in the transfer of knowledge. The impediments to knowledge flow
mentioned above exist in any R&D operations, but they are exacerbated in
global R&D operations. This area of research has received less attention, thus
the second question of this research is:
RESEARCH QUESTION 2: What are the impediments that MNCs

experience in their attempts to disseminate knowledge in their
global R&D operations and how are they overcoming these?
Finally, it is important to consider the “so what?” question, that is whether

knowledge flows have any significant impact on the performance of the R&D
organization. Our basic premise, of course, is that those MNCs that are
successfully facilitating the flow of knowledge throughout their global R&D
operations, and overcoming the impediments, will see a higher level of R&D
performance (see figure 1). Obviously, there are many other factors that affect
R&D performance such as management, availability of resources, strategic
direction, operational effectiveness, competitive environment, etc.
Recognizing this, it is still of interest to compare the observed level of
knowledge flow with the R&D performance in an organization. Measures of
R&D performance might include such things as time-to-market, the rate of new
product introductions, as well as the ability to create radical innovations. This
leads to the following research question. Figure 1 summarizes our conceptual
arguments so far.
RESEARCH QUESTION 3: What impact –if any—does the MNC’s

ability to enhance the knowledge flows in its global R&D operations
have on its R&D performance?
460 ARTICLE SEVEN
Figure 1 Research Model
IT
Applications
Knowledge Flow
R&D
Within R&D
Performance
Operations
Organizational
Measures
METHODOLOGY
As the purpose of this research was to explore the three research questions
above, it was decided to focus on a limited number of MNCs. The selection
criteria were based on a number of factors: 1) annual sales greater than USD 15
billion, 2) large, globally dispersed R&D operations, and 3) operating in the
high technology electronics sector. Three companies were chosen, Hewlett
Packard (HP), one other US company (A), and a Swedish company (B). The
latter two are disguised, according to the wishes of the companies. A brief
description of these companies is provided in table 2.
We adopted a case research approach to the empirical investigation,
because of the importance of studying knowledge flow processes in their real-
life context (Yin, 1989). This approach was particularly important given our
emphasis on studying what actual mechanisms were being employed for
managing knowledge flows, rather than the mechanisms intended for
knowledge management by top management. A secondary reason for choosing
a case study approach was that we felt the existing body of literature did not
adequately describe the phenomenon under investigation. As stated by
Eisenhardt (1989:548), “There are times when little is known about a
phenomenon, current perspectives seem inadequate because they have little
empirical substantiation, or they conflict with each other or common sense. In
these situations, theory building from case study research is particularly
appropriate.”
Table 2 Description of Sample Companies
Company A Company B HP
Headquarter Location United States Sweden United States
1996 Sales USD 17 billion USD 15 billion USD 43 billion
Number of Employees 87,000 93,000 123,000
R&D Operations US, Canada, Located in 23 Extensive
France, United countries in network in
Kingdom, North Asia, Europe,
Japan America, and the
Europe, and Americas
Asia
Number of Interviews
Conducted 25 20 10
At each of these three companies, ten to twenty-five in-depth field

interviews were conducted from June 1997 to February 1998 (see table 2).
People at different levels of the company: corporate R&D manager, laboratory
manager, project manager, and researcher, were interviewed for one-and-a-half
to two hours each. Several steps were taken to increase the reliability and
validity of the results. For example, two interviewers were present at all the
interviews, which were based on a semi-structured interview guide. Also, each
interview was taped and transcribed by one of the interviewers. Immediately
following each interview, individual impressions were discussed by the
interviewers and differences were resolved. Some written material was also
collected from the companies. See appendix for the interview protocol. The
data analysis proceeded through several stages. First the interview data were
reduced and classified based on the R&D matrix. Next the interviews were
analyzed for commonalties based on the research question of challenges in
knowledge dissemination. Three main categories of challenges were
subsequently developed.
Before getting into a description of the findings, it is valuable to provide
a brief discussion of the R&D organizations of the three firms, and thus the
context in which knowledge flows were occurring. All three firms are major
players in the high-technology electronics sector, and this has a number of
implications for how they manage R&D. First, the fast rate of technological
change in this sector has made time-to-market a critical business imperative,
with the result that the development process has become highly structured and
streamlined. Research is also becoming more oriented towards the needs of the
business units, though in HP the research labs are still funded centrally (rather
than through business unit funding). Second, there is an increasing emphasis
on software development in all three companies, as high as 80% of the total
R&D budget in the case of Company B. This creates on the one hand a very
462 ARTICLE SEVEN
fertile environment for knowledge sharing, because of the increasing ease with
which information is transferred over the Internet, and because of the common
tools and languages used by software engineers around the world. However, it
also creates some risks and challenges for knowledge management if a reliance
on software and computer-based interaction drives out face-to-face interaction.
FINDINGS ON KNOWLEDGE DISSEMINATION

MECHANISMS
The knowledge dissemination mechanisms that were most commonly observed
have been mapped onto our classification matrix (see Appendix A for a more
specific discussion of each mechanism).
Table 3 Most Commonly Observed Knowledge Dissemination

Mechanisms
Problem-solving Problem-solving Coordinating
Individual Group Organization
IT Competence Groupware (A) Patent database
Applics. database (HP ) Best practice (A,B,HP)
Communication database (B,HP) Project database
facilitators, e.g., (A,B,HP)
video, email Other databases (HP )
(A,B,HP)
Organ. Personnel rotation Cross-disciplinary Strategic review boards

Measures (A,B,HP) teams (A,B,HP) (A,B,HP)
Interest groups
(A,HP)
Note: A,B,HP represents which company had implemented this measure.
A Focus on Coordinating over Problem-solving Mechanisms by

Management
The majority of mechanisms were created by top-down initiatives to facilitate
knowledge flow for coordination purposes. The primary goal was to overcome
the geographical divisions between the R&D units by encouraging the tracking
of and sharing of knowledge on a management level. The companies all had
well-developed systems for coordination.
However, in terms of the mechanisms to facilitate the flow of
knowledge for the researchers in their everyday problem-solving activities,
there was a wide variety of mechanisms that were implemented in differing
degrees across the companies. There was not as much support from
management for the development of these activities as there was for the
coordinating mechanisms. This seems surprising since mechanisms for
knowledge flow across the units would enable researchers to work more
efficiently and effectively. Of interest, the few mechanisms that had been
implemented by management were not well used by the researchers as
management had developed them from their viewpoint and not from the
individual researcher’s.
Organizational Coordination of Activities
IT Applications
This category primarily included database applications such as databases of
invention proposals, patents, and projects. The drive behind these was from
management who designed them to provide an inventory of the company’s
research activities. These enabled management to conduct such activities as
tracking headcount or the number of patents by discipline. The patent and
project databases were the most well maintained due to management’s mandate
to keep them up-to-date. In several interviews, it was clear that these were
solely for management’s use as access was quite limited in most cases. This is
presumably because while the flow of knowledge within a company is
generally positive, it also increases the ease with which sensitive information
can leak to the outside world.
Company B had an extensive intranet with status and operational
information on each project across the laboratories. As mentioned above, each
project was mandated to enter this information and there was even a handbook
with guidelines for entering the information in the proper way. However, few
steps to facilitate the search for information had been implemented. It was
necessary to search for information laboratory by laboratory, which was a very
slow process with 50 facilities.
Other applications of interest in this category are those developed by
HP. These included a market research database and an external standards
database. The main driver of their creation was to improve the company’s
efficiency. The market research database was designed to create an online
inventory of all market research reports that individuals throughout the
company had obtained. This enabled the company to avoid purchasing more
than one copy of such reports and save time. The second database was
designed with the goal of improving the company’s lobbying efforts with
official standards organizations. It provided an online repository of
information relating to all external standards with listings of the members of
various committees (including if a member of the company was on the
committee).
The measures in this category enabled the companies to coordinate activities of
a more strategic nature, such as funding allocation between the labs or project
review. In all three companies, research strategy reviews were held annually in
464 ARTICLE SEVEN
which the top management from the distributed labs met. This facilitated both
the reductions of cross-disciplinary boundaries and geographic boundaries
while encouraging the sharing of ideas across the labs. Also, despite the extra
cost, all three companies had moved to organize their research by research
theme rather than geographic location as another high-level means to
coordinate the activities across geographic boundaries.
Company A had recently reorganized across five research themes that
crossed geographical boundaries. This was a step in the direction of improving
knowledge flow. However, due to the strong laboratory sub-cultures, the flow
of knowledge still seemed to be hampered. Company B had been organized by
research theme for several years. This had greatly increased the flow of
knowledge and people across borders as people from different geographical
sites had to work together. As mentioned above, HP was in the process of
reorganization during the time of this study and this was a result of the need to
better manage knowledge within the company and to gain better access
knowledge outside the company around the globe.
Another measure of Company A was the establishment of cross-
disciplinary invention proposal review boards. These boards were created for
two reasons. The first reason was because an increasing number of inventions
were cross-disciplinary thus subject experts from several disciplines were
required to review the proposal. The second reason was to bring together
cross-disciplinary experts around invention proposals to facilitate the cross-
pollination of ideas across the labs.
Individual Problem-solving Activities

As mentioned above, the companies had not progressed very far in their efforts
to install mechanisms that facilitated the problem-solving activities of the
researchers either as individuals or within the group. In most cases when the
researchers were asked how they determined whether the company had a
specific competence and where it was located, the most common answer was
that they asked their project manager or the researcher sitting next to him. The
answer was usually found then by email or telephone through a network of
people.
IT Applications
Due to fewer efforts placed on implementing mechanisms to facilitate research,
there were a wide variety of mechanisms and varying levels of use across and
within the companies. These applications could be divided into 1) repositories
of explicit knowledge, 2) links to knowledge, and 3) communication
facilitators. Of the listed IT applications, no company had implemented all of
them. HP was the most advanced in its overall development and use of
applications across labs. However, Company A, which allowed the most
autonomy, had developed the most potentially useful applications for its
researchers. However, they had not been distributed for use across the
organization.
1) Repositories of Explicit Knowledge. As mentioned above, it was clear that

the companies had not designed the invention proposal and patent databases
with the researcher in mind since access to them was quite limited due to the
fear of the flight of intellectual capital from the company. Researchers had to
ask specifically for permission to access the different databases and this
became complicated when asking for access to a database in another
laboratory. However, the companies did have document databases where
researchers could search for published papers on selected topics of interest.
2) Links to Knowledge. Examples of the second group included a competence
or expert database and individual or project webpages. HP had implemented an
elaborate competence database in which any employee could search the entire
company for an employee with a specific competence. In Company B, all
project teams were mandated to create webpages describing their respective
project, yet no search engine that could search across all labs had been
implemented for these. Company A was in the process of developing such an
application; however, it was still in its infancy.
3) Communication Facilitators. The third group consisted of applications such
as email and groupware. E-mail was widely used at all three companies.
Video conferencing was also used at all three companies; however, no
company used it extensively. One manager at Company B felt that the real
take-off of video conferencing would occur when it came to the desktop.
Company A had implemented a new form of groupware that was
developed in-house. This application was among the most interesting of any of
the observed IT applications. However, due to the poor level of integration
among the units, this application appeared to be only well used at one of the
laboratories. This groupware was a virtual workspace, enabling researchers to
easily share information, collaborate on documents, and connect with fellow
researchers through such means as bulletin boards and calendars. Of interest is
that there was no central authority or administration for the system as it was
community owned and maintained. Individual researchers determined their
own level of participation and could designate what level of security they
desired for the information that they entered, e.g., ”write-only” or ”read-only”
access. This seemed to be a well-designed application as the number of users
increased substantially since its introduction.
The organizational measures included such mechanisms as best practice
transfer, process documentation, cross-disciplinary teams, and personnel
rotation, and they were implemented to varying degrees across the companies.
466 ARTICLE SEVEN
Company B had the most structured and hierarchical process for the
transfer of best practice. Once a year, the managers from each R&D laboratory
within the same research area would meet to present to each other what they
felt was best practice. The process of transfer between the laboratories was
voluntary after this. However, if a laboratory manager did not want to adopt
the practice, he or she had to show to management that there was a better way,
which was basically the same as management mandating the transfer.
HP had a well-balanced push-pull attitude to best practice. The
organization had even established a unit that was responsible for the creation
and transfer of best practice in process development across the laboratories.
However, each laboratory was not mandated to use this unit. Each laboratory
had the right to look anywhere in the company as well as outside the company
for ways of improving their processes. Thus, the unit had to market itself to the
rest of the company. Both management and the laboratories seemed satisfied
with this arrangement since it seemed to work well. The creation of this unit
and internal market indicates that the laboratories of HP searched continuously
for best practices, thus creating a pull for knowledge flow.
In line with the best practice transfer, only Company B had a formalized
process for documenting the knowledge gained during a research project. In
fact, no research project was considered completed until this documentation
had occurred. Company A had no such process and could give no examples of
best practice transfer when asked during the interviews. The interviewees
claimed that this was due to the strong subcultures among the labs that led to
the Not-Invented-Here syndrome.
One measure that was both top-down and bottom-up at the companies
was personnel rotation that served to facilitate the transfer of tacit knowledge
across the labs. Company A did have a formal program of personnel rotation;
however, it seemed to be used sparingly. Again due to the subculture rivalry,
individually initiated rotation seemed to be discouraged. One engineer
interviewed said that it took over a year to be transferred to another lab to work
on a project there. He claimed that this was due to the lab managers’ concerns
that it would appear that one lab was being raided by the other and that the lab
losing the person would then have to reduce its headcount by one.
Company B had extensive personnel rotation with most employees
rotating to new locations every 2 to 3 years. Rotation was either requested by
management due to competence needs in another laboratory or by the employee
as part of his or her competence development. In fact, even though rotation
was not articulated as a requirement for advancement in the company, there
was an unwritten policy that this was necessary if one wanted to advance. HP
also had an extensive personnel rotation scheme and as one interviewee said,
“There are 12 different ways to rotate at this company.”
Another means to facilitate knowledge transfer was to use researchers
from across several laboratories in one project. Company B used this
extensively as all of its projects involved more than one laboratory on a project.
Work was even reshuffled throughout the laboratories as demands for different
levels of competence varied across projects. Company A was involved in some
cross-laboratory projects although the majority of projects were within each
laboratory. HP had initiated a virtual R&D laboratory project two years prior.
Unfortunately, it was not possible to gather any data on this project at the time.
In addition to the above, there were some grass-roots efforts such as
technology conferences or technology interest groups. The goal of these efforts
was to bring together people with similar interests and competencies from
across the organization. The level of encouragement from management
differed across the companies. HP fully encouraged these conferences and
groups and provided resources for these activities. One of the purposes of the
technology interest groups was to provide funding for back-burner research as
well as to promote the creation of networks across the company. These efforts
worked very well and participation within the interest groups was high.
Company A did not discourage such groups; however, no resources were
allocated to enable such efforts. This was mirrored in the low participation
rates. Company B did not have any such grass-roots group. However, one
interviewee at Company B stated that his unit felt that there was a real need for
such cross-border groups. Although no sponsoring was given for internal
groups, management did sponsor the creation of groups between a laboratory
and its environment. In such groups one or two researchers joined two to six
people from outside the company in the same location who had the same
expertise. When interesting ideas developed, the researchers made a formal
presentation to the management of the laboratory who then decided on any
actions to be taken.
Identifying and Overcoming Impediments to Knowledge

Dissemination
The second aim of this research was to pay explicit attention to the obstacles to
knowledge flows within the global operations. The findings of interest
concerned the knowledge flows occurring at the level of the researcher and not
at the level of the organization. This is primarily because management focused
on implementing and mandating the use of these mechanisms. Since less
attention had been placed on the individual researcher’s needs for problem-
solving activities, considerable impediments were found to exist to cross-
laboratory knowledge flows. The main challenges observed were of a
behavioral nature, implying that even if all the structural channels existed,
knowledge would still not flow due to the employees’ lack of willingness to
disseminate or assimilate knowledge. A discussion of these impediments as
well as the means to help overcome these is provided below.
468 ARTICLE SEVEN
Table 4 Impediments to the Dissemination of Knowledge

Individual and Group Organizational
Problem-solving Coordination
IT Applications Opportunity cost of time Keeping up-to-date
Linking databases
geographically
Linking IP with patent
with project
Organizational Knowledge is power Strategic agreement across

Measures Not-invented-here laboratories
Organizational Coordination
For the IT applications which facilitated management’s tracking of research
activities, the underlying challenge was to ensure that up-to-date information
was entered and that the databases across the labs were linked. In addition,
links had to be made across the different types of databases. For example, when
an invention proposal was patented, then this should be entered in the invention
proposal database. In all companies, administrative procedures had been
implemented to ensure information input into the databases, thus overcoming
any challenge to keeping them up-to-date. Interestingly this was a considerable
challenge at HP due to the frequent movement of its employees.
With respect to the organizational measures, the main challenge was
more on a strategic level. In other words, with respect to the review boards,
management at each laboratory had to agree to the overall strategic direction of
the laboratories and the allocation of resources among the laboratories. In
addition, significant coordinating challenges are raised when the laboratories
are organized based on research theme as opposed to geographic location;
however, these challenges become more operational as opposed to the transfer
of knowledge.
Problem-solving Activities
The more interesting challenges from this paper’s perspective arose when
discussing the mechanisms to facilitate the everyday activities of researchers.
In an ideal knowledge-flowing world, researchers would have instantaneous
access to all information and know-how dispersed throughout the company.
However, even if an infrastructure were implemented, several impediments to
the flow of knowledge would still be present. Three major categories were
observed in the analysis: Opportunity Cost of Time, Knowledge is Power, and
Not-Invented-Here. In this study, the individual researchers were seen as both
givers and receivers of knowledge, and opportunity cost of time was a
challenge on both dimensions. However, Knowledge is Power was seen to be a
challenge on the giver side while Not-Invented-Here was a challenge on the

receiver side.
Opportunity Cost of Time. The successful implementation of the IT

applications is partly dependent upon having up-to-date and relevant
information. When discussing with the interviewees the implementation of
these systems, it was often stated that they had not entered their own
information because it was too time consuming, they were unsure of what
benefit they would receive for doing so, and they did not want to be bothered
with telephone calls from people finding them in the database.
In order to surmount these challenges, several different ideas were
developed. One idea was to create a nag feature that prompted the user every
time he or she logged onto the system. A message appeared stating how many
days it had been since the information had last been updated. Another idea was
to encourage the use of the system through bonus features. At one company
when the user entered the appropriate data for the project database, an
automatic email distribution list of project members was created. For another
system, updating information was rewarded with a coffee coupon. For those
concerned with being contacted too often for their expertise, a feature was
added which allowed them to specify by which means they were to be
contacted or during what times, e.g., by email only.
Opportunity Cost of Time was also mentioned when discussing
organizational measures, however, to a lesser degree. Attending technology
conferences or participating in technology interest groups also might be
constrained by time. Participation in these was low in Company A due to the
lack of support from management. On the other hand, HP had the most active
cross-geography technology interest groups of the three. This seemed to be due
to the fact that management showed its support for these groups and that they
were of a grass-roots nature. People who were interested in a technology or a
subject were free to attend those that were of interest.
Knowledge is Power and Not-Invented-Here (NIH). Knowledge is Power and

Not-Invented-Here seem to be opposite sides of the same coin. A researcher
does not want to reveal certain knowledge as he or she feels that this ensures a
certain amount of power. On the other hand, a researcher may not want to
assimilate someone else’s knowledge as he or she might feel that this would
represent lesser intelligence. However, it was important for researchers to get
over these feelings and work together since new inventions at Company A, for
example, normally required at least two inventors due to the increasingly
complex technical nature of the product. In fact, the company had an average
of 2.3 inventors per patent. HP seemed to have a completely different attitude
among researchers that mirrored the general philosophy that the company
would take an idea from anywhere as long as it was a good one. This
470 ARTICLE SEVEN
contrasted with Company A that had a history of developing everything in-

house. At Company B researchers seemed open to new ideas from others and
open to sharing. However, Company B was more open to ideas from within
the organization than from outside.
One observed means to overcome these knowledge flow barriers was to
base the researcher’s performance appraisal not only on individual performance
but also on the individual’s contribution to the knowledge of the company.
This practice is quite common in management consultancies where an “up-or-
out” policy often reinforces self-interest, thus the addition of knowledge
sharing criteria helps overcome knowledge sharing challenges. In Company A,
the performance appraisals did not include any contributions to the company’s
knowledge base. One researcher who had made a big attempt to facilitate the
cross-pollination of ideas across labs did not feel that there was much response
from management concerning this effort during his performance review.
Company B took a different approach as it encouraged personal development in
their performance appraisals; however, the company did not consider the
individual’s contribution to the company’s knowledge base. HP had a created a
push-pull system in the manner it structured its performance appraisal. Half of
an individual’s performance review was based on individual results while the
other half was based on how the results were achieved, i.e., the efficient use of
resources (pull) and the contribution back to the company (push).
In addition to the above, management actively attempted to show its
support for knowledge sharing activities. Company A with the least amount of
interlab knowledge dissemination suffered from a lack of management support.
Ideas that crossed laboratory boundaries received little support from
management. In addition in Company A, technology councils which were
formed to bring together those with similar technology interests were poorly
supported by management thus the level of commitment and participation was
low. At HP management stood firmly behind such interest groups, providing
financial resources as well as encouraging participation in such groups.
Moving away from the level of the individual, Knowledge is Power or
NIH was also observed at the level of the project team or even of the lab. In
Company A, the different laboratories were resistant to ideas from the other
laboratories due to the strong subcultures observed. Also, the laboratories
seemed to be concerned that sharing ideas between laboratories might affect the
allocation of resources between the laboratories. Both Company B and C had
laboratories that were more open to new ideas that seemed to stem partly from
the setting of goals by each company. Company B established the goal of each
of the laboratories reaching the level 5 of the “Capability Maturity Model”.
Under this model, the laboratories conducted quality audits of each other to
determine the level of each laboratory. This system encouraged the searching
for ideas by the individual laboratories and mirrored the aspect of personal
development on the individual performance appraisals at the company. HP had
established the goal of becoming the “World’s Best Industrial Labs” for all of
the laboratories as a whole. This also facilitated openness and sharing between
the laboratories.
DISCUSSION AND CONCLUSIONS

From the above findings, successful knowledge dissemination in global R&D
operations appears to be dependent on a two-part knowledge management
system. The first part of this system includes a supporting infrastructure of IT
applications and organizational measures. However, even if all the support and
infrastructure are in place, there is no guarantee that knowledge will flow
smoothly and efficiently between the laboratories due to the challenges that
were identified above, especially at the individual level. Thus, management
must grease the wheels of knowledge flow through the implementation of
behavioral mechanisms that help create a one-company culture.
Supporting Infrastructure
The underlying element of the knowledge management system is a supporting
infrastructure of both IT applications as well as organizational measures that
facilitate the transfer of knowledge. IT applications for the repository of
knowledge such as a competence database are useful for finding articulated
knowledge or links to the knowledge source in an efficient manner.
In order to ensure the efficiency of these databases, our research
suggests that they must be up-to-date as information changes rapidly. This is
not an easy task due to the impediment of the individual’s opportunity cost of
time. The means to achieve up-to-date entries ranged from management
mandates to a voluntary basis. From the findings on databases, it seems that
management has to consider the tradeoff of mandating the upkeep of its
databases that requires policing or administrative resources or allowing it to be
voluntary which does not require resources.
Company A’s groupware that created a virtual community workspace to
coordinate activities and facilitate communication was a very interesting
application and it was surprising to see that it was not implemented in all of the
company’s laboratories. This was most likely due to many subcultures within
the company. This type of groupware was not seen at the other companies.
The use of this groupware was not mandated and researchers contributed of
their own free will. The person contributing information had control over the
use of these documents by being able to limit who had access to the documents
or whether the document was for read-only or write-only. Participants seemed
to respond positively to this self-control.
Another means of facilitating the spread of knowledge was through the
implementation of a best practice database. It would be expected that within
R&D operations that the knowledge gained during a project would be
472 ARTICLE SEVEN
documented so that future projects could gain from this knowledge. Both
Company B and HP had developed such databases. Company B had a strict
policy, mandating this documentation at the end of all projects. Company A
claimed that since projects never really ended, there was never any clear time
to document the knowledge gained; however, this seemed to be an easy way
out of performing a time-consuming activity.
One challenge with documenting project knowledge is that the
researchers have difficulty determining what knowledge is valuable to others as
well as codifying it so others understand. HP facilitated the transfer of best
practice through the creation of a unit whose task was to ensure that knowledge
was captured and transferred successfully. This unit sent people to the different
laboratories to facilitate the transfer of the more tacit knowledge as well.
Even though the companies are spending significant resources on the
creation and implementation of IT applications, it was observed that
researchers still only communicated largely with those whom they already
knew. This may be because the systems were not very well developed or too
difficult to use. In Company A one researcher (who was one of the star
researchers) showed a real lack of interest in the company intranet. When
asked if he used it, he said that he only did so for communication through email
and to find out about company policies such as vacation days. He claimed that
the intranet did not provide any ”high-bandwidth information” and that it was
more effective to speak with one of his fellow researchers sharing offices next
to his. This supports the research on communities of practice. When
attempting to solve a problem, a researcher asks someone else within his or her
community of practice for help. This person may be next door or across the
globe. A common language and trust within the community have already been
established which thus facilitates the flow of knowledge.
Thus, management acknowledged the need to encourage the interaction
of people who work on similar problems, especially those who would not
normally meet, to facilitate knowledge flow. Discussion of ideas as well as the
narrating of stories encourages the pull of knowledge from one laboratory to
another. This feat is difficult to achieve through any IT application so the best
means of achieving this are the organizational measures discussed above, e.g.,
personnel rotation, global conferences on specific themes, technology interest
groups, wired coffee rooms, and cross-functional project teams. Even
arranging the offices in an open layout encourages communication. Again HP
had implemented many of these features while at the same time providing both
financial and management support.
One-company Culture
Organizational culture is the set of values and resulting practices, concerning
relationships among people and the world around them, that is shared by people
in an organization. Further, as Meyerson (1991:256) noted, “Organizational
culture is the code word for the subjective side of organizational life.“
Organizational culture is difficult to measure and therefore traditionally
quantitative methods have been avoided as organizational culture is normally
best studied through qualitative methods (e.g., observation and interviews)
(Schein, 1985). While we certainly only touch upon investigating case-study
firms’ organizational cultures in this study, by following Schein’s
recommendations we feel that we have been able to obtain a reasonable
assessment of the firms’ organizational cultures through the interviews we
conducted and through observations made while visiting the case study firms.
A central issue that emerged as an important determinant of knowledge
flow was the extent to which a company had established a one-company culture
as opposed to having several sub-cultures. Building the supporting
infrastructure creates the channels for knowledge flow; however, behavioral
challenges can restrict the knowledge flow through these channels. Company
A was characterized by strong sub-cultures and thus had the least amount of
knowledge flow across the laboratories of the three companies. For example,
interviewees made comments on a number of occasions about how differently
things were done at the company’s different geographical locations. The
interviewees also appeared to feel more pride in their laboratory affiliation than
their company affiliation. Also, significant differences in the physical design,
dress, and style of people were clearly apparent to the researchers at the
different labs. There was little pull for knowledge as there seemed to be a
sense of rivalry among the laboratories. This also led to the lack of push of
knowledge.
Company B had succeeded in building somewhat of a one-company
culture through a high level of personnel rotation. However, there seemed
primarily to be pull for knowledge within the operations due to the personal
development aspect of the performance appraisal and the internal rivalry of
reaching level 5 in the capability maturity model. Thus, the company had a
better degree of knowledge flow than Company A.
HP had built a one-company culture to the greatest degree of the three
companies and perhaps as a result, it had the greatest flow of knowledge within
its global operations. The HP culture encouraged the sharing of ideas with
others as well as taking the time to help others. When asked why there was
such a sharing culture, several interviewees mentioned that this was partly due
to the core values of the company, one of which was teamwork. One
interviewee stated that if she were to stop at someone’s desk anywhere in the
company to ask for help, she felt that 99.9% of the time, the employee would
take the time to help. Another means which HP used was the company-wide
goal of becoming the “World’s Best Industrial Labs”. Management
encouraged the generation of ideas to improve the R&D operations in any
manner, e.g., process improvements, environmental concerns, etc., through
providing resources to the best ideas. This goal seemed to work better than the
474 ARTICLE SEVEN
goal set by Company B since the ideas presented in HP often were

improvements that spanned laboratory boundaries as opposed to improvements
for each lab.
Referring to the community of practice literature, these findings support
the discussion concerning identity (Wenger, 1998). In HP a strong common
identity among employees across the global operations had been built.
Employees felt that their identity was tied to the company as a whole and not
just to their division, thus creating a balance between firm identity and division
identity. This enabled the acceptance of ideas that were from outside the lab as
they were not seen to be threatening to the identity of the lab. In addition, the
one-company culture encouraged the contribution of ideas as employees were
interested in improving the whole of the company. Both self-interest and
community-interest were overcome as the betterment of the entire company
meant the betterment of each employee and laboratory.
Company A, on the other hand, seemed to suffer from divisions with
which employees identified more with their laboratory than with the company
as a whole. This led to the feeling of fiefdoms within the company, thus
laboratory interest or community interest was stronger than interest in the
whole company. As mentioned, this was exhibited when discussing personnel
rotation between the laboratories. Managers were afraid of losing someone to
another laboratory, which might mean a reduction in headcount and resources.
In addition, there was resistance to ideas coming from outside the division due
to the subcultures. This was clear as each division had developed its own
knowledge sharing tools for within the division; however, these had not been
adapted by the other divisions even though they were aware of them. In
addition, employees seemed to have a higher degree of self-interest than those
at HP. They exhibited more concern with their time as well as a higher degree
of Knowledge is Power and Not-Invented-Here.
Figure 2 Overcoming Knowledge Flow Impediments
Knowledge Flows
Challenges to Knowledge Flow

•Opportunity Cost of Time
•Knowledge is Power
•Not-Invented-Here
Self-interest
Self-interest
Self-interest
Community Interest Self-interest
Self-interest
Creating a One-Company Culture

•Core Value of Teamwork Self-interest
•Corporate-level goals
•Knowledge-sharing appraisals Community Interest
From these observations, a one-company culture helps reduce the

behavioral impediments to knowledge flow. This one-company culture can be
encouraged through several means: instilling a feeling of teamwork as one of
the core values of the company, overcoming community interest through goals
aimed at the improvement of company as a whole, and overcoming self-interest
through adapting performance appraisals to reflect an individual’s contribution
to the company’s knowledge base. In addition, a high level of personnel
rotation can be seen to contribute to a one-company culture through the
building of a common language and shared values and cross-border networks.
Linking to R&D Performance

It was proposed at the beginning of the paper that a high degree of knowledge
flow within global R&D operations would be linked to high R&D performance.
To examine whether this was the case, we administered a brief questionnaire to
interview respondents to get their perceptions of their company’s performance.
The company averages are listed in table 5.
Interviews suggested that Company A had the poorest internal R&D
knowledge flow, and this is consistent with their relatively weak ratings on
time to market, revenues from products developed in the last three years, and
impact of R&D on the emergence of successful products. While we know from
the interviews that Company A has historically been very good at developing
radical innovations, they have clearly struggled to commercialize these
innovations, which at least in part seems to be a function of their poor internal
knowledge flow. Interestingly, Company A scored the highest on articles
published per technical employee, which is consistent with their apparent focus
on idea creation, rather than commercialization.
Company B had a considerably better knowledge flow than Company A.
It scored very well on time to market, and impact of R&D on emergence of
successful products, but it scored lowest on end customer satisfaction and
articles published per technical employee. These numbers suggest a very
impressive R&D performance, but one that is achieved through focusing
knowledge flows on one thing (getting products to market fast) with a possible
loss of concern for building longer-term effectiveness.
HP had the most balanced performance assessment, rating almost as high as
company B on the emergence of successful products, highest on customer
satisfaction, and highest on quality of R&D work. On no measures did they
score the lowest. These findings follow logically from our assessment of HP as
having the most effective systems for managing internal knowledge flows. The
usual caveats regarding self-rated performance and subjective indexes apply,
but the performance data is at least consistent with our interpretation of the
internal knowledge flows in the three companies.
476 ARTICLE SEVEN
Table 5 Knowledge Management Mechanisms and Performance
Co. A Co. B HP
IT Applications
Coordinating
Patent database 1 2 2
Project database 1 2 1
Problem-Solving
Best practice database -- 1 2
Competence database -- -- 1
Project web pages 1 2 1
Groupware 1 -- --
Email 2 2 2
Video conferencing 1 1 1
Coordinating
Cross-lab review boards 2 2 2
Problem-solving
Personnel rotation 1 3 2
Technology interest groups 1 -- 3
Cross-disciplinary teams 2 1 1
Behavioral Impediments
Opportunity cost of time 2 1 1
Knowledge is power 2 1 1
Not-invented-here 2 1 1
One-Company Culture Measures
Teamwork in company values -- -- 2
Performance appraisal – knowledge dev. -- 1 2
Company goals established -- 1 2
Overall Knowledge Flow
Intralab 3 2 3
Interlab 1 2 3
R&D Performance Measures
Impact of R&D on emergence of successful 2.9 3.8 3.3
products (B > A)
% Revenue from products developed in the last 2.8 4.0 3.7
three years (B > A, HP > A)
Speed, time to market (B>A, HP > A) 2.0 2.7 2.7
End customer satisfaction (HP > B) 3.3 2.9 3.8
Quality of R&D work (HP > A) 3.0 3.3 4.0
Articles published per technical employee (A > B) 3.6 2.7 3.4
- Note for knowledge flow performance measures: Numbers refer to researcher’s subjective
assessments where: -- Non-existent, 1 – Poor performance, 2 – Good performance, 3 – Excellent
performance.
- Note for R&D performance measures: Each person interviewed was asked to complete a very brief
questionnaire. They were asked: “Estimate the performance of your division over the last three years,
in comparison to competitor firms in your industry where 1= much worse and 5= much better”.
Responses listed in the above table are average responses from each firm. Parentheses indicate which
pairs of numbers are statistically different at 0.05 level using the Kruskal-Wallis ANOVA test.
Finding the Balance

Both building the IT infrastructure as well as encouraging interaction through
organizational measures require significant resources. Finding the right
balance between the two is a significant challenge for management. However,
the most important component in facilitating knowledge transfer among global
R&D operations is to create a knowledge sharing culture that promotes
curiosity for and openness to new ideas. This then creates a pull for knowledge
which management can support through creating channels for knowledge flow.
A push or desire for knowledge contribution is also an important factor in
creating successful knowledge flows.
Like all studies, this study has some limitations. We have explicitly
focused on the mechanisms of knowledge transfer in this paper. Future
research would benefit from a deeper exploration of innovation processes. In
addition, we have chosen a deep exploration of several case study firms to
obtain rich detail and because of the exploratory nature of this study. This
obviously inhibits our ability to generalize, which is left for future large sample
research.
From the companies observed in this research, it appears that
management has primarily focused on developing and implementing
knowledge management flow systems aimed at fulfilling their coordination
activities. Less attention has been paid to the individual researcher; however,
this is the main resource of a company’s R&D operations. Management should
be more interested in implementing measures that would enable the researcher
to work more efficiently while at the same time more innovatively. Thus, one
future research direction would be to focus on better understanding the
individual researcher, actually studying the way in which the researcher works
and the means that the researchers uses to obtain knowledge for problem-
solving activities. What knowledge does the researcher need? Where is it
located and how does the researcher access it? What are researchers’
preferences for individual control, job security, and reward structure, and how
do they in turn affect knowledge dissemination? What can firms do to align
individual and group goals in R&D organizations? Such research would enable
management to better understand the requirements for successful knowledge
transfer mechanisms and thus, enable companies to improve knowledge flow
within their global R&D operations.
REFERENCES
Allen, T.J. 1977. Managing the Flow of Technology. Cambridge: MIT Press.
Asakawa, K. 1995. Managing Knowledge Conversion Process across Borders:
Toward a Framework of International Knowledge Management, working
paper, Insead.
478 ARTICLE SEVEN
Barth, G. et al. 1995. Managing International Research & Development

Networks, working paper, Carnegie Bosch Institute.
Bartlett, C.A. & Ghoshal, S. 1989. Managing across Borders: The
Behrman, J.N. & Fisher, W.A. 1980. Overseas R&D Activities of Transnational
Companies. Cambridge: Oelgeschlager, Gunn and Hain.
Behrman, J.N. & Fisher, W.A. 1980. Transnational Corporations: Market
Orientations and R&D Abroad. Columbia Journal of World Business, 15, 3:
55-60.
Blakeslee, S. 1985. Clues Hint at Brain’s Two Memory Maps. The New York
Times, Feb. 19, C1.
Brown, J.S. & Duguid, P. 1991. Organizational Learning and Communities-of-
Practice: Toward a Unified View of Working, Learning, and Innovation,
Organization Science, 2,1: 1991.
Davenport, T. & Prusak, L. 1998. Working Knowledge, Harvard Business
School Press.
Doz, Y., Asakawa, K., Santos, J.E.P., & Williamson, P.J. 1997. The
Metanational Corporation. Working paper, Insead.
DeMeyer, A. 1989. Global R&D Management. R&D Management, 19,2.
DeMeyer, A. 1991. Tech Talk: How Managers are Stimulating Global R&D
Communication. Sloan Management Review, Spring.
Eisenhardt, K. 1989. Building Theories from Case Study Research. Academy of
Granstrand, O., Håkanson, L., & Sjölander, S. 1992 Technology Management
and International Business - Internationalization of R&D and Technology.
Chichester: Wiley.
Grant, R.M. 1997. The Knowledge-Based View of the Firm: Implications for
Management Practice. Long-Range Planning, 30,3: 450-454.
Håkanson, L. 1990. International Decentralization of R&D – The
Organizational Challenges, in Bartlett, C.A., Doz, Y. & Hedlund, G. (eds.),
Managing the Global Firm, London: Routledge.
Håkanson, L. & Zander, U. 1986. Managing International Research &
Development, Mekanpublikation.
Hedlund, G. 1994. A Model of Knowledge Management and the N-form
Corporation, Strategic Management Journal, 15.
Hedlund, G. & Nonaka, I. 1993. Models of Knowledge Management in the
West and Japan in Lorange, P., Chakravarthy, B., Roos, J. & Van de Van, A.
(eds.), Implementing Strategic Processes: Change, Learning, &
Cooperation. Basil Blackwell.
Hedlund, G. & Zander, U. 1992. Architectronic and List-like Structuring: A
Critique of Modern Concepts of Management, working paper, Institute of
International Business.
Kogut, B. & Zander, U. 1992. Knowledge of the Firm, Combinative
Capabilities, and the Replication of Technology. Organization Science, 3,3:
383-397.
Kuemmerle, W. 1997. Building Effective R&D Capabilities Abroad. Harvard

Business Review, March-April.
Lall, S. 1979. The International Allocation of Research Activity by US
multinationals, Oxford Bulletin of Economics and Statistics, 41: 313-331.
Mansfield, E., Teece, D. & Romeo, A. 1979. Overseas Research and
Development by U.S. Based Firms. Economica, 46: 187-196.
Meyerson, D. & Martin, J. Acknowledging and uncovering ambiguities, in
Frost, P., Moore, L., Louis, M., Lundberg, C. & Martin, J. (eds.), Reframing
Organizational Culture, Beverly Hills: Sage.
Nobel, R. & Birkinshaw, J. 1998. Innovation in Multinational Corporations:
Control and Communication Patterns in International R&D Operations.
Strategic Management Journal, 19: 479-496.
Nonaka, I. & Takeuchi, H. 1995. The Knowledge-Creating Company: How
Japanese Companies Create the Dynamics of Innovation. Oxford University
Press.
Pasmore, W.A. & Gurley, K. 1991. Enhancing R&D across Functional Areas.
In R.H. Kilmann (ed.), Making Organizations Competitive. Jossey-Bass:
368-398.
Polanyi, M. 1969. Knowing and Being, London: Routledge and Kegan Paul.
Ronstadt, R. C. 1977. Research and Development Abroad by U.S.
Multinationals, New York: Praeger Publishers.
Schein, E. 1985. Organizational Culture and Leadership. San Francisco:
Jossey-Bass.
Snyder, W. 1996. Organization Learning and Performance: An Exploration of
the Linkages between Organization Learning, Knowledge, and Performance.
unpublished doctoral dissertation, University of Southern California.
Snyder, W. 1997. Communities-of-Practice: Combining Organization Learning
and Strategy Insights to Create a Bridge to the 21rst Century, presented at
the Academy of Management.
Szulanski, G. 1996. Exploring Internal Stickiness: Impediments to the Transfer
of Best Practice Within the Firm. Strategic Management Journal, 17: 27-43.
von Hippel, E. 1988. Sources of Innovation, Oxford University Press.
Terpstra, V. 1997. International Product Policy: The Role of Foreign R&D.
Columbia Journal of World Business, Winter: 24-32.
Wenger, E. 1998. Communities-of-Practice: Learning, Meaning, and Identity,
Westney, E. 1993. Cross-Pacific Internationalization of R&D by US and
Japanese Firms. R&D Management, 23,2: 171-181.
Yin, R.K. 1989. Case Study Research- Design and Methods. London: Sage
Publications.
Zander, U. 1991. Exploiting a Technological Edge – Voluntary and Involuntary
Dissemination of Technology. Published doctoral dissertation, Institute of
International Business.
480 ARTICLE SEVEN
APPENDIX A – OVERVIEW OF KNOWLEDGE DISSEMINATION MECHANISMS

Coordinating Purpose Challenges to Successful Implementation Co.
IT Applications
Invention Proposal To create online inventory of submitted invention proposals Fear of flight of intellectual property, defining who has access A
Database Keeping up-to-date: 1) adding new proposals, 2) connection to patent
Patent Database To create online inventory of all patents obtained by company Keeping up-to-date: 1) adding new patents A,B,C
Linking across laboratories
Project Database To create online inventory of all current projects, descriptions, and participants Defining who has access A,B,C
Motivating people to enter information and keep up-to-date
Market Research To create an online inventory of all market research reports obtained by the Motivating people to keep up-to-date C
Database company
External Standards To create an online repository of all information relating to external standards - Motivating people to keep up-to-date C
Database company members on standards committees
Org. Measures
Cross-Disciplinary To create board of cross-disciplinary experts to evaluate invention proposals Ensuring interest from the different laboratories A
IP Review Boards To facilitate the diffusion of ideas across laboratories Ensuring that panels include representatives from different sites
Cross-Disciplinary To create cross-disciplinary forum for discussion of overall research direction Ensuring consensus over direction among the laboratories A,B,C
Strategy Reviews To facilitate the diffusion of ideas across laboratories
Problem-Solving Purpose Challenges to Successful Implementation Co.

IT Applications
Idea Board To facilitate the connection of ideas while still in the formulation phase Fear of others stealing ideas A
Degree of formal structure
Employee To create an online database of employees and their competencies Definition of competence A
Competence Motivating people to enter information – lack of time and interest
Database Keeping up with employee movements and competence development
Guide to Databases To create an online listing of all personal databases Definition of database – personal rolodex vs. research test results C
Project Webpages To give overview of each project by laboratory Motivating people to keep up-to-date A,B,C
Employee To facilitate the spread of ideas and tools Motivating people to enter information C
Webpages
Invention Proposal To provide researchers with the status of invention proposals – date of review, Defining who has access and to what A
Review Groupware feedback, decision, etc.
Community To create bulletin board and calendar Defining who has access and to what A
Groupware To facilitate the sharing of documents Motivating people to participate
Library Groupware To facilitate the communication between librarians and library research analysts Motivating people to participate C
Org. Measures
Project documenting To document knowledge gained in research project Motivating people to take the time B,C
Best Practice To promote the transfer of best practices across the laboratories Not-invented-here syndrome B,C
Processes Motivating people to take the time to contribute
Technology To bring together researchers with common interests Encouraging people to participate A,B,C
Conferences To access knowledge from external speakers Securing resources from management
Problem-Solving Purpose Challenges to Successful Implementation Co.

Technology Interest To create the sense of communication across the laboratories Encouraging commitment A,C
Groups Securing resources from management
Library Research To assign one researcher to each research group Ensuring communication among research analysts C
Analysis Liaisons To facilitate knowledge flow across research groups through liaisons
Personnel Rotation To facilitate the transfer of know-how across research laboratories Providing incentives and support for researchers to move A,B,C
Cross-Laboratory To bring together people from across the laboratories Coordination and communication between laboratories A,B,C
Teams To help create one-company culture
482 ARTICLE SEVEN
APPENDIX B
Knowledge Management - Challenges in R&D Organizations
Interview Protocol
Questions Directed to Corporate Management / R&D Center Directors

1. Please describe your company’s R&D organization. In particular,
- Where are the main units located, and what are their primary
responsibilities?
- How much interaction is there between R&D units, e.g., in terms of
frequency of visits, exchanges of personnel, video conferencing, email,
faxes, etc.
- To what extent are these different units’ activities coordinated? E.g., do
you have any joint projects between units?
- How many people are involved in a typical research project?
2. To what extent do you see the following aspects of knowledge management
being undertaken in your company’s R&D organization:
- Knowledge transfer, i.e., from one unit to another
- Knowledge acquisition, i.e., where one unit seeks out and gets hold of
knowledge from other places
- Knowledge dissemination, i.e., where one unit makes knowledge
available to multiple other units
- Knowledge combination / creation, i.e., where two or more units get
together and create something new
3. What aspects of knowledge are most relevant in your R&D organization?
Please indicate the extent to which the following types of knowledge are
relevant:
- “Intellectual property”, e.g., patents, blueprints, proprietary
technologies
- “Public domain” knowledge, e.g., about industry standards, technical
questions
- Process management technology, e.g., how you make things, how you
do things
- Market knowledge, e.g., what customer needs are in various locations
- Knowledge about specific people or contacts
4. What are the major obstacles to knowledge management in your company’s
R&D organization? Specifically, to what extent is each of the following a
problem:
- A secretive culture, people “hoarding” knowledge, lack of trust?
- A lack of time?
- Difficulty in making knowledge explicit?

- The Not-Invented-Here syndrome?
- A lack of awareness of what knowledge is out there?
- A difficulty in retrieving and adapting knowledge?
5. What systems does your company use to facilitate the transfer of knowledge
within its R&D operations?
- IT systems, e.g., intranets, knowledge databases?
- For your IT systems, what information is included in these? Who has
access to these? Who is responsible for updating these?
- “Structural solutions”, e.g., committees, centers of excellence, cross-
functional teams, knowledge brokers, financial compensation for
transferring knowledge?
- Informal networks, e.g., “communities of practice”, socialization
mechanisms?
- Formalized activities, e.g., conferences, internal memos? Who attends
these or receives these?
- What incentives exist to promote the use of these systems?
- What barriers do you see to the effective use of these systems?
6. How are the researchers in your R&D units evaluated? How often is this
evaluation?
7. Please describe a recent case that you are aware of where knowledge was
effectively transferred from one R&D unit to another.
8. Do you think the knowledge management in your company’s R&D
organization is effective? How could it be improved?
484 ARTICLE SEVEN
Questions Directed to R&D Project Managers

Please answer the following questions keeping in mind how knowledge
management was used in one particular R&D project you are/were involved
with.
1. Please identify a specific R&D project that you have worked with/work on
(ideally one that has been in existence for some time).
2. To what extent has your R&D project undertaken the following aspects of
knowledge management:
- Knowledge transfer, i.e., from one R&D project to another
- Knowledge acquisition, i.e., where one R&D project seeks out and gets
hold of knowledge from other places
- Knowledge dissemination, i.e., where one R&D project makes
knowledge available to multiple other R&D projects
- Knowledge combination / creation, i.e., where two or more R&D
projects get together and create something new
3. What aspects of knowledge have been most useful to your R&D project?
Please indicate the extent to which the following types of knowledge have been
useful:
- “Intellectual property”, e.g., patents, blueprints, proprietary technologies
- “Public domain” knowledge, e.g., about industry standards, technical
questions
- Process management technology, e.g., how you make things, how you
do things
- Market knowledge, e.g., what customers needs are in various locations
- Knowledge about specific people or contacts
4. What are the major obstacles to knowledge management that people
involved in your R&D project have faced? Specifically, to what extent is
each of the following a problem:
- A secretive culture, people “hoarding” knowledge, lack of trust?
- A lack of time?
- Difficulty in making knowledge explicit?
- The Not-Invented-Here syndrome?
- A lack of awareness of what knowledge is out there?
- A difficulty in retrieving and adapting knowledge?
5. What systems have been used by your R&D project to facilitate the transfer
of knowledge?
- IT systems, e.g., intranets, knowledge databases?
- For your IT systems, what information is included in these? Who has
access to these? Who is responsible for updating these?
- “Structural solutions”, e.g., committees, centers of excellence, cross-

functional teams, knowledge brokers, financial compensation for
transferring knowledge?
- Informal networks, e.g., “communities of practice”, socialization
mechanisms? How do these get started? Who is involved in these?
- Formalized activities, e.g., conferences, internal memos? Who attends
these or receives these? What is the frequency of these?
- What incentives exist to promote the use of these systems?
- What barriers do you see to the effective use of these systems?
6. How are your researchers evaluated? How often is this evaluation?
7. If you are interested in learning more about a certain area within your
company’s R&D operations, how do you proceed?
8. Please describe a recent example when your R&D project was involved in
transferring knowledge from one location to another.
9. Do you think the knowledge management in your company’s R&D
organization is effective? How could it be improved?
Published Doctoral Dissertations from IIB
1 Leksell, Laurent: Headquarters-Subsidiary Relationships in

Multinational Corporations. (Out of print)
Stockholm: IIB/EFI, 1981. ISBN 91-7258-133-6.
2 Spångberg, Kjell: Strategi i diversifierade företag -
huvudkontorets roll. (Out of print)
Stockholm: IIB/EFI, 1982. ISBN 91-7258-137-9.
3 Lindgren, Ulf: Foreign Acquisitions - Management of the
Integration Process. (Out of print)
Stockholm: IIB/EFI, 1982. ISBN 91-7258-145-X.
4 Forss, Kim: Planning and Evaluation in Aid Organizations.
Stockholm: IIB/EFI, 1985. ISBN 91-7258-196-4. SEK 800.-.
5 Sölvell, Örjan: Entry barriers and foreign penetration - Emerging
patterns of international competition in two electrical engineering
industries.
Stockholm: IIB, 1987. ISBN 91-7810-999-X. SEK 800.-.
6 Roos, Johan: Cooperative Venture Formation Processes:
Characteristics and Impact on Performance. (Out of print)
Stockholm: IIB, 1989. ISBN 91-971005-3-6.
7 Ågren, Lars: Swedish Direct Investment in the U.S.
Stockholm: IIB, 1990. ISBN 91-971005-4-4. SEK 800.-.
8 Åhlander, Karl: Aspects of Modern Treasury Management -
Organization and external financial activities in Swedish MNCs.
Stockholm: IIB, 1990. ISBN 91-971005-95. SEK 800.-.
9 Nordström, Kjell A: The Internationalization Process of the Firm -
Searching for New Patterns and Explanations.
10 Berglöf, Erik: Corporate Control and Capital Structure - Essays on
Property Rights and Financial Contracts.
11 Zander, Udo: Exploiting A Technological Edge - Voluntary and
Involuntary Dissemination of Technology.
12 Lindqvist, Maria: Infant Multinationals - The Internationalization
of Young, Technology-Based Swedish Firms.
13 Hagström, Peter: The 'Wired' MNC. The Role of Information
Systems for Structural Change in Complex Organizations.
14 Zander, Ivo: The Tortoise Evolution of the Multinational
Corporation - Foreign Technological Activity in Swedish
Multinational Firms 1890 - 1990.
488 ARTICLE SEVEN

15 Ridderstråle, Jonas: Global Innovation - Managing International
Innovation Projects at ABB and Electrolux.
16 Zander, Lena: The Licence to Lead - An 18 Country Study of the
Relationship Between Employees’ Preferences Regarding
Interpersonal Leadership and National Culture.
17 Denrell, Jerker: Essays on the Economic Effects of Vanity and
Career Concerns.
18 Arvidsson, Niklas: The Ignorant MNE – The Role of Perception
Gaps in Knowledge Management.
19 Regnér, Patrick: Strategy Creation and Change in Complexity –
Adaptive and Creative Learning Dynamics in the Firm.
Stockholm: IIB, 1999. ISBN 91-971730-7X. SEK 500.-.
20 Jonsson, Stefan: Making and Breaking Norms - Competitive
Imitation Patterns in the Swedish Mutual Fund Industry
Stockholm, IIB, 2002. ISBN 91-971730-96. SEK 500.-.
21 Åman, Per: Revolution by Evolution - Transforming international
management in the established MNC
Stockholm, IIB, 2003. ISBN 91-971730-10. SEK 500.-.
22 Lerpold, Lin: Reputation by Association – Exploring alliance
formation and organizational identity adaptation
Stockholm, IIB, 2003. ISBN 91-973849-0-9. SEK 500:-.
23 Teigland, Robin: Knowledge Networking – Structure and
Performance in Networks of Practice
Stockholm, IIB, 2003. ISBN 91-973849-1-7. SEK 500:-.
The recently developed knowledge-based view of the firm argues that knowledge is the
firm’s most valuable resource, yet as managers are discovering, valuable firm knowledge can
easily leak across the firm’s legal boundaries just as it can get stuck within them. Research
on work practices has consistently shown that social relationships are an important factor
in these knowledge flows (or lack of them). However, the twist is that individuals generally
form these relationships not according to what the formal organization dictates, but based
on personal biases and preferences for others who may be either inside or outside the firm.
These relationships are the basis for informal, naturally occurring networks or “networks
of practice”. A tension then arises because much of the knowledge within organizations
is controlled by individuals who make discretionary choices about the sharing of their
knowledge, and as a result, management is finding that knowledge cannot be “managed”
using the same tools that once were appropriate for dealing with physical goods.
The overarching goal of this thesis is to improve our understanding of networks of practice
from a business firm’s perspective, and in particular to investigate issues of structure and
performance, two important areas generally left by the wayside in previous network of
practice research. The study incorporates seven empirical studies in organizations including
business firms, such as the more traditional Cap Gemini and Hewlett-Packard and the
“less” traditional internet-consulting multinational, Icon Medialab; a large infrastructure
project led by Skanska; and an online discussion community, consisting of U.S. lawyers
nationwide. Using concepts, such as collective action, public goods, weak ties, and
knowledge integration, and methods including a social network analysis of 1698 individuals
across 16 countries, surveys, and case studies, this thesis supports taking a differentiated
view of networks of practice over a unitary one. For example, our results reveal difficulty in
applying a common set of structural properties to all types of networks of practice and that
levels of efficient and creative performance vary depending on the network of practice.
Thus, this research contributes to an expanding inter-disciplinary field that examines the
creation and sharing of knowledge in organizations as a source of competitive advantage.
The results have theoretical implications for research fields such as networks of practice,
social networks, and the knowledge-based view of the firm as well as empirical implications
for managers striving to encourage knowledge creation and sharing within their firms.
ISBN 91-973849-1-7
www.teigland.com

Teigland Thesis-Knowledge Networking

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Teigland Thesis-Knowledge Networking

Uploaded by

Copyright:

Available Formats

KNOWLEDGE NETWORKING

Structure and Performance in Networks of Practice

Structure and Performance

• Was founded in 1975

The IIB Board of Directors

Professor Leif Lindmark, President, Stockholm School of Economics

Associate Professor Peter Hagström, Director IIB

More information: www.sse.edu/iib

Structure and Performance

Stockholm School of Economics

© IIB and the author

ARTICLE TWO ................................................................................................... 305

ARTICLE THREE ................................................................................................ 319

ARTICLE FOUR .................................................................................................. 349

ARTICLE FIVE ................................................................................................... 363

ARTICLE SIX...................................................................................................... 393

ARTICLE SEVEN ................................................................................................ 453

Figure 1.1 Networks of Practice as a Subset of Other Networks......................... 6

Table 3.1 Selected Studies of Communities of Practice .................................... 65

In addition to my committee, I would like to especially thank two people,

Several companies and individuals have played a particularly important

A few others deserve mentioning. While I always pondered getting a

Finally, I would like to acknowledge my family. I am a very fortunate

On a final note, it is impossible to express what I owe those closest to me

Stocksund, August 20, 2003

“..….but most importantly I have my network from the

“People from other offices [within the company] call me up

THE RECENTLY DEVELOPED knowledge-based view of the firm argues

While organizational theorists have long discussed the importance of emergent

addition, an important rationale for studying emergent networks lies in the

networks from a business firm’s perspective. Before discussing how we intend

1.2 Defining Networks of Practice

to describe the organizationally specified role relationships between individuals

As a result and contrary to traditional views, emergent relationships that are

definition of networks of practice, we see networks of practice as a subset of all

Figure 1.1 Networks of Practice as a Subset of Other Networks

1.3 Overview of Research Purposes and Empirical Studies

In short, our first step is to conduct an extensive literature review to determine

1.3.1 Developing an Understanding of Networks of Practice

geographic location. The second dimension refers to the primary

Figure 1.2 Matrix of Networks of Practice

Primary Communication Channels

Face-to-face Mixed Electronic

1.3.2 Extending Our Understanding through Empirical Studies

1.3.2.1 Research Purpose 1: Structural Dimensions of Networks of Practice

Research Purpose 1: To describe the structural properties of networks

1.3.2.2 Research Purpose 2: Performance and Networks of Practice

However, within the network of practice literature, there is a surprising scarcity

Additionally, as we will find in our literature review, the majority of the

tends to consider the already existing constellation of individuals, taking a point

been made and individual job performance. As such, Research Purpose 2

Research Purpose 2: To investigate the relationship between

To address Research Purpose 2, we conduct a series of studies at two

1.3.3 Summary of Research Purposes

1.4 Definitions and Delimitations

The focus of this study is on networks of practice and in particular on structure

noted above, we define practice as an activity, an interaction among individuals

In our second research purpose, we move from structure to performance, and in

A second major area of debate revolves around the individual/collective

recent sociological approach sees knowledge as socially constructed and

As for delimitations, despite the strengths of this thesis being a multi-level,

Regarding the first research purpose focusing on network of practice structure,

individuals had access to numerous networks of practice through various

1.5 Preview of the Study and Major Contributions